Which of the following functions would you use to find the column location of specified text

Matching / broadcasting behavior#

DataFrame has the methods add(), sub(), mul(), div() and related functions radd(), rsub(), … for carrying out binary operations. For broadcasting behavior, Series input is of primary interest. Using these functions, you can use to either match on the index or columns via the axis keyword:

In [18]: df = pd.DataFrame( ....: { ....: "one": pd.Series(np.random.randn(3), index=["a", "b", "c"]), ....: "two": pd.Series(np.random.randn(4), index=["a", "b", "c", "d"]), ....: "three": pd.Series(np.random.randn(3), index=["b", "c", "d"]), ....: } ....: ) ....: In [19]: df Out[19]: one two three a 1.394981 1.772517 NaN b 0.343054 1.912123 -0.050390 c 0.695246 1.478369 1.227435 d NaN 0.279344 -0.613172 In [20]: row = df.iloc[1] In [21]: column = df["two"] In [22]: df.sub(row, axis="columns") Out[22]: one two three a 1.051928 -0.139606 NaN b 0.000000 0.000000 0.000000 c 0.352192 -0.433754 1.277825 d NaN -1.632779 -0.562782 In [23]: df.sub(row, axis=1) Out[23]: one two three a 1.051928 -0.139606 NaN b 0.000000 0.000000 0.000000 c 0.352192 -0.433754 1.277825 d NaN -1.632779 -0.562782 In [24]: df.sub(column, axis="index") Out[24]: one two three a -0.377535 0.0 NaN b -1.569069 0.0 -1.962513 c -0.783123 0.0 -0.250933 d NaN 0.0 -0.892516 In [25]: df.sub(column, axis=0) Out[25]: one two three a -0.377535 0.0 NaN b -1.569069 0.0 -1.962513 c -0.783123 0.0 -0.250933 d NaN 0.0 -0.892516

Furthermore you can align a level of a MultiIndexed DataFrame with a Series.

In [26]: dfmi = df.copy() In [27]: dfmi.index = pd.MultiIndex.from_tuples( ....: [(1, "a"), (1, "b"), (1, "c"), (2, "a")], names=["first", "second"] ....: ) ....: In [28]: dfmi.sub(column, axis=0, level="second") Out[28]: one two three first second 1 a -0.377535 0.000000 NaN b -1.569069 0.000000 -1.962513 c -0.783123 0.000000 -0.250933 2 a NaN -1.493173 -2.385688

Series and Index also support the divmod() builtin. This function takes the floor division and modulo operation at the same time returning a two-tuple of the same type as the left hand side. For example:

In [29]: s = pd.Series(np.arange(10)) In [30]: s Out[30]: 0 0 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 dtype: int64 In [31]: div, rem = divmod(s, 3) In [32]: div Out[32]: 0 0 1 0 2 0 3 1 4 1 5 1 6 2 7 2 8 2 9 3 dtype: int64 In [33]: rem Out[33]: 0 0 1 1 2 2 3 0 4 1 5 2 6 0 7 1 8 2 9 0 dtype: int64 In [34]: idx = pd.Index(np.arange(10)) In [35]: idx Out[35]: Int64Index([0, 1, 2, 3, 4, 5, 6, 7, 8, 9], dtype='int64') In [36]: div, rem = divmod(idx, 3) In [37]: div Out[37]: Int64Index([0, 0, 0, 1, 1, 1, 2, 2, 2, 3], dtype='int64') In [38]: rem Out[38]: Int64Index([0, 1, 2, 0, 1, 2, 0, 1, 2, 0], dtype='int64')

We can also do elementwise divmod():

In [39]: div, rem = divmod(s, [2, 2, 3, 3, 4, 4, 5, 5, 6, 6]) In [40]: div Out[40]: 0 0 1 0 2 0 3 1 4 1 5 1 6 1 7 1 8 1 9 1 dtype: int64 In [41]: rem Out[41]: 0 0 1 1 2 2 3 0 4 0 5 1 6 1 7 2 8 2 9 3 dtype: int64

Missing data / operations with fill values#

In Series and DataFrame, the arithmetic functions have the option of inputting a fill_value, namely a value to substitute when at most one of the values at a location are missing. For example, when adding two DataFrame objects, you may wish to treat NaN as 0 unless both DataFrames are missing that value, in which case the result will be NaN (you can later replace NaN with some other value using fillna if you wish).

In [42]: df Out[42]: one two three a 1.394981 1.772517 NaN b 0.343054 1.912123 -0.050390 c 0.695246 1.478369 1.227435 d NaN 0.279344 -0.613172 In [43]: df2 Out[43]: one two three a 1.394981 1.772517 1.000000 b 0.343054 1.912123 -0.050390 c 0.695246 1.478369 1.227435 d NaN 0.279344 -0.613172 In [44]: df + df2 Out[44]: one two three a 2.789963 3.545034 NaN b 0.686107 3.824246 -0.100780 c 1.390491 2.956737 2.454870 d NaN 0.558688 -1.226343 In [45]: df.add(df2, fill_value=0) Out[45]: one two three a 2.789963 3.545034 1.000000 b 0.686107 3.824246 -0.100780 c 1.390491 2.956737 2.454870 d NaN 0.558688 -1.226343

Flexible comparisons#

Series and DataFrame have the binary comparison methods eq, ne, lt, gt, le, and ge whose behavior is analogous to the binary arithmetic operations described above:

In [46]: df.gt(df2) Out[46]: one two three a False False False b False False False c False False False d False False False In [47]: df2.ne(df) Out[47]: one two three a False False True b False False False c False False False d True False False

These operations produce a pandas object of the same type as the left-hand-side input that is of dtype bool. These boolean objects can be used in indexing operations, see the section on Boolean indexing.

Boolean reductions#

You can apply the reductions: empty, any(), all(), and bool() to provide a way to summarize a boolean result.

In [48]: (df > 0).all() Out[48]: one False two True three False dtype: bool In [49]: (df > 0).any() Out[49]: one True two True three True dtype: bool

You can reduce to a final boolean value.

In [50]: (df > 0).any().any() Out[50]: True

You can test if a pandas object is empty, via the empty property.

In [51]: df.empty Out[51]: False In [52]: pd.DataFrame(columns=list("ABC")).empty Out[52]: True

To evaluate single-element pandas objects in a boolean context, use the method bool():

In [53]: pd.Series([True]).bool() Out[53]: True In [54]: pd.Series([False]).bool() Out[54]: False In [55]: pd.DataFrame([[True]]).bool() Out[55]: True In [56]: pd.DataFrame([[False]]).bool() Out[56]: False

Warning

You might be tempted to do the following:

Or

These will both raise errors, as you are trying to compare multiple values.:

ValueError: The truth value of an array is ambiguous. Use a.empty, a.any() or a.all().

See gotchas for a more detailed discussion.

Comparing if objects are equivalent#

Often you may find that there is more than one way to compute the same result. As a simple example, consider df + df and df * 2. To test that these two computations produce the same result, given the tools shown above, you might imagine using (df + df == df * 2).all(). But in fact, this expression is False:

In [57]: df + df == df * 2 Out[57]: one two three a True True False b True True True c True True True d False True True In [58]: (df + df == df * 2).all() Out[58]: one False two True three False dtype: bool

Notice that the boolean DataFrame df + df == df * 2 contains some False values! This is because NaNs do not compare as equals:

In [59]: np.nan == np.nan Out[59]: False

So, NDFrames (such as Series and DataFrames) have an equals() method for testing equality, with NaNs in corresponding locations treated as equal.

In [60]: (df + df).equals(df * 2) Out[60]: True

Note that the Series or DataFrame index needs to be in the same order for equality to be True:

In [61]: df1 = pd.DataFrame({"col": ["foo", 0, np.nan]}) In [62]: df2 = pd.DataFrame({"col": [np.nan, 0, "foo"]}, index=[2, 1, 0]) In [63]: df1.equals(df2) Out[63]: False In [64]: df1.equals(df2.sort_index()) Out[64]: True

Comparing array-like objects#

You can conveniently perform element-wise comparisons when comparing a pandas data structure with a scalar value:

In [65]: pd.Series(["foo", "bar", "baz"]) == "foo" Out[65]: 0 True 1 False 2 False dtype: bool In [66]: pd.Index(["foo", "bar", "baz"]) == "foo" Out[66]: array([ True, False, False])

pandas also handles element-wise comparisons between different array-like objects of the same length:

In [67]: pd.Series(["foo", "bar", "baz"]) == pd.Index(["foo", "bar", "qux"]) Out[67]: 0 True 1 True 2 False dtype: bool In [68]: pd.Series(["foo", "bar", "baz"]) == np.array(["foo", "bar", "qux"]) Out[68]: 0 True 1 True 2 False dtype: bool

Trying to compare Index or Series objects of different lengths will raise a ValueError:

In [55]: pd.Series(['foo', 'bar', 'baz']) == pd.Series(['foo', 'bar']) ValueError: Series lengths must match to compare In [56]: pd.Series(['foo', 'bar', 'baz']) == pd.Series(['foo']) ValueError: Series lengths must match to compare

Note that this is different from the NumPy behavior where a comparison can be broadcast:

In [69]: np.array([1, 2, 3]) == np.array([2]) Out[69]: array([False, True, False])

or it can return False if broadcasting can not be done:

In [70]: np.array([1, 2, 3]) == np.array([1, 2]) Out[70]: False

Combining overlapping data sets#

A problem occasionally arising is the combination of two similar data sets where values in one are preferred over the other. An example would be two data series representing a particular economic indicator where one is considered to be of “higher quality”. However, the lower quality series might extend further back in history or have more complete data coverage. As such, we would like to combine two DataFrame objects where missing values in one DataFrame are conditionally filled with like-labeled values from the other DataFrame. The function implementing this operation is combine_first(), which we illustrate:

In [71]: df1 = pd.DataFrame( ....: {"A": [1.0, np.nan, 3.0, 5.0, np.nan], "B": [np.nan, 2.0, 3.0, np.nan, 6.0]} ....: ) ....: In [72]: df2 = pd.DataFrame( ....: { ....: "A": [5.0, 2.0, 4.0, np.nan, 3.0, 7.0], ....: "B": [np.nan, np.nan, 3.0, 4.0, 6.0, 8.0], ....: } ....: ) ....: In [73]: df1 Out[73]: A B 0 1.0 NaN 1 NaN 2.0 2 3.0 3.0 3 5.0 NaN 4 NaN 6.0 In [74]: df2 Out[74]: A B 0 5.0 NaN 1 2.0 NaN 2 4.0 3.0 3 NaN 4.0 4 3.0 6.0 5 7.0 8.0 In [75]: df1.combine_first(df2) Out[75]: A B 0 1.0 NaN 1 2.0 2.0 2 3.0 3.0 3 5.0 4.0 4 3.0 6.0 5 7.0 8.0

General DataFrame combine#

The combine_first() method above calls the more general DataFrame.combine(). This method takes another DataFrame and a combiner function, aligns the input DataFrame and then passes the combiner function pairs of Series (i.e., columns whose names are the same).

So, for instance, to reproduce combine_first() as above:

In [76]: def combiner(x, y): ....: return np.where(pd.isna(x), y, x) ....: In [77]: df1.combine(df2, combiner) Out[77]: A B 0 1.0 NaN 1 2.0 2.0 2 3.0 3.0 3 5.0 4.0 4 3.0 6.0 5 7.0 8.0

Summarizing data: describe#

There is a convenient describe() function which computes a variety of summary statistics about a Series or the columns of a DataFrame (excluding NAs of course):

In [94]: series = pd.Series(np.random.randn(1000)) In [95]: series[::2] = np.nan In [96]: series.describe() Out[96]: count 500.000000 mean -0.021292 std 1.015906 min -2.683763 25% -0.699070 50% -0.069718 75% 0.714483 max 3.160915 dtype: float64 In [97]: frame = pd.DataFrame(np.random.randn(1000, 5), columns=["a", "b", "c", "d", "e"]) In [98]: frame.iloc[::2] = np.nan In [99]: frame.describe() Out[99]: a b c d e count 500.000000 500.000000 500.000000 500.000000 500.000000 mean 0.033387 0.030045 -0.043719 -0.051686 0.005979 std 1.017152 0.978743 1.025270 1.015988 1.006695 min -3.000951 -2.637901 -3.303099 -3.159200 -3.188821 25% -0.647623 -0.576449 -0.712369 -0.691338 -0.691115 50% 0.047578 -0.021499 -0.023888 -0.032652 -0.025363 75% 0.729907 0.775880 0.618896 0.670047 0.649748 max 2.740139 2.752332 3.004229 2.728702 3.240991

You can select specific percentiles to include in the output:

In [100]: series.describe(percentiles=[0.05, 0.25, 0.75, 0.95]) Out[100]: count 500.000000 mean -0.021292 std 1.015906 min -2.683763 5% -1.645423 25% -0.699070 50% -0.069718 75% 0.714483 95% 1.711409 max 3.160915 dtype: float64

By default, the median is always included.

For a non-numerical Series object, describe() will give a simple summary of the number of unique values and most frequently occurring values:

In [101]: s = pd.Series(["a", "a", "b", "b", "a", "a", np.nan, "c", "d", "a"]) In [102]: s.describe() Out[102]: count 9 unique 4 top a freq 5 dtype: object

Note that on a mixed-type DataFrame object, describe() will restrict the summary to include only numerical columns or, if none are, only categorical columns:

In [103]: frame = pd.DataFrame({"a": ["Yes", "Yes", "No", "No"], "b": range(4)}) In [104]: frame.describe() Out[104]: b count 4.000000 mean 1.500000 std 1.290994 min 0.000000 25% 0.750000 50% 1.500000 75% 2.250000 max 3.000000

This behavior can be controlled by providing a list of types as include/exclude arguments. The special value all can also be used:

In [105]: frame.describe(include=["object"]) Out[105]: a count 4 unique 2 top Yes freq 2 In [106]: frame.describe(include=["number"]) Out[106]: b count 4.000000 mean 1.500000 std 1.290994 min 0.000000 25% 0.750000 50% 1.500000 75% 2.250000 max 3.000000 In [107]: frame.describe(include="all") Out[107]: a b count 4 4.000000 unique 2 NaN top Yes NaN freq 2 NaN mean NaN 1.500000 std NaN 1.290994 min NaN 0.000000 25% NaN 0.750000 50% NaN 1.500000 75% NaN 2.250000 max NaN 3.000000

That feature relies on select_dtypes. Refer to there for details about accepted inputs.

Index of min/max values#

The idxmin() and idxmax() functions on Series and DataFrame compute the index labels with the minimum and maximum corresponding values:

In [108]: s1 = pd.Series(np.random.randn(5)) In [109]: s1 Out[109]: 0 1.118076 1 -0.352051 2 -1.242883 3 -1.277155 4 -0.641184 dtype: float64 In [110]: s1.idxmin(), s1.idxmax() Out[110]: (3, 0) In [111]: df1 = pd.DataFrame(np.random.randn(5, 3), columns=["A", "B", "C"]) In [112]: df1 Out[112]: A B C 0 -0.327863 -0.946180 -0.137570 1 -0.186235 -0.257213 -0.486567 2 -0.507027 -0.871259 -0.111110 3 2.000339 -2.430505 0.089759 4 -0.321434 -0.033695 0.096271 In [113]: df1.idxmin(axis=0) Out[113]: A 2 B 3 C 1 dtype: int64 In [114]: df1.idxmax(axis=1) Out[114]: 0 C 1 A 2 C 3 A 4 C dtype: object

When there are multiple rows (or columns) matching the minimum or maximum value, idxmin() and idxmax() return the first matching index:

In [115]: df3 = pd.DataFrame([2, 1, 1, 3, np.nan], columns=["A"], index=list("edcba")) In [116]: df3 Out[116]: A e 2.0 d 1.0 c 1.0 b 3.0 a NaN In [117]: df3["A"].idxmin() Out[117]: 'd'

Note

idxmin and idxmax are called argmin and argmax in NumPy.

Value counts (histogramming) / mode#

The value_counts() Series method and top-level function computes a histogram of a 1D array of values. It can also be used as a function on regular arrays:

In [118]: data = np.random.randint(0, 7, size=50) In [119]: data Out[119]: array([6, 6, 2, 3, 5, 3, 2, 5, 4, 5, 4, 3, 4, 5, 0, 2, 0, 4, 2, 0, 3, 2, 2, 5, 6, 5, 3, 4, 6, 4, 3, 5, 6, 4, 3, 6, 2, 6, 6, 2, 3, 4, 2, 1, 6, 2, 6, 1, 5, 4]) In [120]: s = pd.Series(data) In [121]: s.value_counts() Out[121]: 6 10 2 10 4 9 3 8 5 8 0 3 1 2 dtype: int64 In [122]: pd.value_counts(data) Out[122]: 6 10 2 10 4 9 3 8 5 8 0 3 1 2 dtype: int64

New in version 1.1.0.

The value_counts() method can be used to count combinations across multiple columns. By default all columns are used but a subset can be selected using the subset argument.

In [123]: data = {"a": [1, 2, 3, 4], "b": ["x", "x", "y", "y"]} In [124]: frame = pd.DataFrame(data) In [125]: frame.value_counts() Out[125]: a b 1 x 1 2 x 1 3 y 1 4 y 1 dtype: int64

Similarly, you can get the most frequently occurring value(s), i.e. the mode, of the values in a Series or DataFrame:

In [126]: s5 = pd.Series([1, 1, 3, 3, 3, 5, 5, 7, 7, 7]) In [127]: s5.mode() Out[127]: 0 3 1 7 dtype: int64 In [128]: df5 = pd.DataFrame( .....: { .....: "A": np.random.randint(0, 7, size=50), .....: "B": np.random.randint(-10, 15, size=50), .....: } .....: ) .....: In [129]: df5.mode() Out[129]: A B 0 1.0 -9 1 NaN 10 2 NaN 13

Discretization and quantiling#

Continuous values can be discretized using the cut() (bins based on values) and qcut() (bins based on sample quantiles) functions:

In [130]: arr = np.random.randn(20) In [131]: factor = pd.cut(arr, 4) In [132]: factor Out[132]: [(-0.251, 0.464], (-0.968, -0.251], (0.464, 1.179], (-0.251, 0.464], (-0.968, -0.251], ..., (-0.251, 0.464], (-0.968, -0.251], (-0.968, -0.251], (-0.968, -0.251], (-0.968, -0.251]] Length: 20 Categories (4, interval[float64, right]): [(-0.968, -0.251] < (-0.251, 0.464] < (0.464, 1.179] < (1.179, 1.893]] In [133]: factor = pd.cut(arr, [-5, -1, 0, 1, 5]) In [134]: factor Out[134]: [(0, 1], (-1, 0], (0, 1], (0, 1], (-1, 0], ..., (-1, 0], (-1, 0], (-1, 0], (-1, 0], (-1, 0]] Length: 20 Categories (4, interval[int64, right]): [(-5, -1] < (-1, 0] < (0, 1] < (1, 5]]

qcut() computes sample quantiles. For example, we could slice up some normally distributed data into equal-size quartiles like so:

In [135]: arr = np.random.randn(30) In [136]: factor = pd.qcut(arr, [0, 0.25, 0.5, 0.75, 1]) In [137]: factor Out[137]: [(0.569, 1.184], (-2.278, -0.301], (-2.278, -0.301], (0.569, 1.184], (0.569, 1.184], ..., (-0.301, 0.569], (1.184, 2.346], (1.184, 2.346], (-0.301, 0.569], (-2.278, -0.301]] Length: 30 Categories (4, interval[float64, right]): [(-2.278, -0.301] < (-0.301, 0.569] < (0.569, 1.184] < (1.184, 2.346]] In [138]: pd.value_counts(factor) Out[138]: (-2.278, -0.301] 8 (1.184, 2.346] 8 (-0.301, 0.569] 7 (0.569, 1.184] 7 dtype: int64

We can also pass infinite values to define the bins:

In [139]: arr = np.random.randn(20) In [140]: factor = pd.cut(arr, [-np.inf, 0, np.inf]) In [141]: factor Out[141]: [(-inf, 0.0], (0.0, inf], (0.0, inf], (-inf, 0.0], (-inf, 0.0], ..., (-inf, 0.0], (-inf, 0.0], (-inf, 0.0], (0.0, inf], (0.0, inf]] Length: 20 Categories (2, interval[float64, right]): [(-inf, 0.0] < (0.0, inf]]

Tablewise function application#

DataFrames and Series can be passed into functions. However, if the function needs to be called in a chain, consider using the pipe() method.

First some setup:

In [142]: def extract_city_name(df): .....: """ .....: Chicago, IL -> Chicago for city_name column .....: """ .....: df["city_name"] = df["city_and_code"].str.split(",").str.get(0) .....: return df .....: In [143]: def add_country_name(df, country_name=None): .....: """ .....: Chicago -> Chicago-US for city_name column .....: """ .....: col = "city_name" .....: df["city_and_country"] = df[col] + country_name .....: return df .....: In [144]: df_p = pd.DataFrame({"city_and_code": ["Chicago, IL"]})

extract_city_name and add_country_name are functions taking and returning DataFrames.

Now compare the following:

In [145]: add_country_name(extract_city_name(df_p), country_name="US") Out[145]: city_and_code city_name city_and_country 0 Chicago, IL Chicago ChicagoUS

Is equivalent to:

In [146]: df_p.pipe(extract_city_name).pipe(add_country_name, country_name="US") Out[146]: city_and_code city_name city_and_country 0 Chicago, IL Chicago ChicagoUS

pandas encourages the second style, which is known as method chaining. pipe makes it easy to use your own or another library’s functions in method chains, alongside pandas’ methods.

In the example above, the functions extract_city_name and add_country_name each expected a DataFrame as the first positional argument. What if the function you wish to apply takes its data as, say, the second argument? In this case, provide pipe with a tuple of (callable, data_keyword). .pipe will route the DataFrame to the argument specified in the tuple.

For example, we can fit a regression using statsmodels. Their API expects a formula first and a DataFrame as the second argument, data. We pass in the function, keyword pair (sm.ols, 'data') to pipe:

In [147]: import statsmodels.formula.api as sm In [148]: bb = pd.read_csv("data/baseball.csv", index_col="id") In [149]: ( .....: bb.query("h > 0") .....: .assign(ln_h=lambda df: np.log(df.h)) .....: .pipe((sm.ols, "data"), "hr ~ ln_h + year + g + C(lg)") .....: .fit() .....: .summary() .....: ) .....: Out[149]: <class 'statsmodels.iolib.summary.Summary'> """ OLS Regression Results ============================================================================== Dep. Variable: hr R-squared: 0.685 Model: OLS Adj. R-squared: 0.665 Method: Least Squares F-statistic: 34.28 Date: Tue, 22 Nov 2022 Prob (F-statistic): 3.48e-15 Time: 05:34:17 Log-Likelihood: -205.92 No. Observations: 68 AIC: 421.8 Df Residuals: 63 BIC: 432.9 Df Model: 4 Covariance Type: nonrobust =============================================================================== coef std err t P>|t| [0.025 0.975] ------------------------------------------------------------------------------- Intercept -8484.7720 4664.146 -1.819 0.074 -1.78e+04 835.780 C(lg)[T.NL] -2.2736 1.325 -1.716 0.091 -4.922 0.375 ln_h -1.3542 0.875 -1.547 0.127 -3.103 0.395 year 4.2277 2.324 1.819 0.074 -0.417 8.872 g 0.1841 0.029 6.258 0.000 0.125 0.243 ============================================================================== Omnibus: 10.875 Durbin-Watson: 1.999 Prob(Omnibus): 0.004 Jarque-Bera (JB): 17.298 Skew: 0.537 Prob(JB): 0.000175 Kurtosis: 5.225 Cond. No. 1.49e+07 ============================================================================== Notes: [1] Standard Errors assume that the covariance matrix of the errors is correctly specified. [2] The condition number is large, 1.49e+07. This might indicate that there are strong multicollinearity or other numerical problems. """

The pipe method is inspired by unix pipes and more recently dplyr and magrittr, which have introduced the popular (%>%) (read pipe) operator for R. The implementation of pipe here is quite clean and feels right at home in Python. We encourage you to view the source code of pipe().

Row or column-wise function application#

Arbitrary functions can be applied along the axes of a DataFrame using the apply() method, which, like the descriptive statistics methods, takes an optional axis argument:

In [150]: df.apply(np.mean) Out[150]: one 0.811094 two 1.360588 three 0.187958 dtype: float64 In [151]: df.apply(np.mean, axis=1) Out[151]: a 1.583749 b 0.734929 c 1.133683 d -0.166914 dtype: float64 In [152]: df.apply(lambda x: x.max() - x.min()) Out[152]: one 1.051928 two 1.632779 three 1.840607 dtype: float64 In [153]: df.apply(np.cumsum) Out[153]: one two three a 1.394981 1.772517 NaN b 1.738035 3.684640 -0.050390 c 2.433281 5.163008 1.177045 d NaN 5.442353 0.563873 In [154]: df.apply(np.exp) Out[154]: one two three a 4.034899 5.885648 NaN b 1.409244 6.767440 0.950858 c 2.004201 4.385785 3.412466 d NaN 1.322262 0.541630

The apply() method will also dispatch on a string method name.

In [155]: df.apply("mean") Out[155]: one 0.811094 two 1.360588 three 0.187958 dtype: float64 In [156]: df.apply("mean", axis=1) Out[156]: a 1.583749 b 0.734929 c 1.133683 d -0.166914 dtype: float64

The return type of the function passed to apply() affects the type of the final output from DataFrame.apply for the default behaviour:

• If the applied function returns a Series, the final output is a DataFrame. The columns match the index of the Series returned by the applied function.

• If the applied function returns any other type, the final output is a Series.

This default behaviour can be overridden using the result_type, which accepts three options: reduce, broadcast, and expand. These will determine how list-likes return values expand (or not) to a DataFrame.

apply() combined with some cleverness can be used to answer many questions about a data set. For example, suppose we wanted to extract the date where the maximum value for each column occurred:

In [157]: tsdf = pd.DataFrame( .....: np.random.randn(1000, 3), .....: columns=["A", "B", "C"], .....: index=pd.date_range("1/1/2000", periods=1000), .....: ) .....: In [158]: tsdf.apply(lambda x: x.idxmax()) Out[158]: A 2000-08-06 B 2001-01-18 C 2001-07-18 dtype: datetime64[ns]

You may also pass additional arguments and keyword arguments to the apply() method. For instance, consider the following function you would like to apply:

def subtract_and_divide(x, sub, divide=1): return (x - sub) / divide

You may then apply this function as follows:

df.apply(subtract_and_divide, args=(5,), divide=3)

Another useful feature is the ability to pass Series methods to carry out some Series operation on each column or row:

In [159]: tsdf Out[159]: A B C 2000-01-01 -0.158131 -0.232466 0.321604 2000-01-02 -1.810340 -3.105758 0.433834 2000-01-03 -1.209847 -1.156793 -0.136794 2000-01-04 NaN NaN NaN 2000-01-05 NaN NaN NaN 2000-01-06 NaN NaN NaN 2000-01-07 NaN NaN NaN 2000-01-08 -0.653602 0.178875 1.008298 2000-01-09 1.007996 0.462824 0.254472 2000-01-10 0.307473 0.600337 1.643950 In [160]: tsdf.apply(pd.Series.interpolate) Out[160]: A B C 2000-01-01 -0.158131 -0.232466 0.321604 2000-01-02 -1.810340 -3.105758 0.433834 2000-01-03 -1.209847 -1.156793 -0.136794 2000-01-04 -1.098598 -0.889659 0.092225 2000-01-05 -0.987349 -0.622526 0.321243 2000-01-06 -0.876100 -0.355392 0.550262 2000-01-07 -0.764851 -0.088259 0.779280 2000-01-08 -0.653602 0.178875 1.008298 2000-01-09 1.007996 0.462824 0.254472 2000-01-10 0.307473 0.600337 1.643950

Finally, apply() takes an argument raw which is False by default, which converts each row or column into a Series before applying the function. When set to True, the passed function will instead receive an ndarray object, which has positive performance implications if you do not need the indexing functionality.

Aggregation API#

The aggregation API allows one to express possibly multiple aggregation operations in a single concise way. This API is similar across pandas objects, see groupby API, the window API, and the resample API. The entry point for aggregation is DataFrame.aggregate(), or the alias DataFrame.agg().

We will use a similar starting frame from above:

In [161]: tsdf = pd.DataFrame( .....: np.random.randn(10, 3), .....: columns=["A", "B", "C"], .....: index=pd.date_range("1/1/2000", periods=10), .....: ) .....: In [162]: tsdf.iloc[3:7] = np.nan In [163]: tsdf Out[163]: A B C 2000-01-01 1.257606 1.004194 0.167574 2000-01-02 -0.749892 0.288112 -0.757304 2000-01-03 -0.207550 -0.298599 0.116018 2000-01-04 NaN NaN NaN 2000-01-05 NaN NaN NaN 2000-01-06 NaN NaN NaN 2000-01-07 NaN NaN NaN 2000-01-08 0.814347 -0.257623 0.869226 2000-01-09 -0.250663 -1.206601 0.896839 2000-01-10 2.169758 -1.333363 0.283157

Using a single function is equivalent to apply(). You can also pass named methods as strings. These will return a Series of the aggregated output:

In [164]: tsdf.agg(np.sum) Out[164]: A 3.033606 B -1.803879 C 1.575510 dtype: float64 In [165]: tsdf.agg("sum") Out[165]: A 3.033606 B -1.803879 C 1.575510 dtype: float64 # these are equivalent to a ``.sum()`` because we are aggregating # on a single function In [166]: tsdf.sum() Out[166]: A 3.033606 B -1.803879 C 1.575510 dtype: float64

Single aggregations on a Series this will return a scalar value:

In [167]: tsdf["A"].agg("sum") Out[167]: 3.033606102414146

Transform API#

The transform() method returns an object that is indexed the same (same size) as the original. This API allows you to provide multiple operations at the same time rather than one-by-one. Its API is quite similar to the .agg API.

We create a frame similar to the one used in the above sections.

In [185]: tsdf = pd.DataFrame( .....: np.random.randn(10, 3), .....: columns=["A", "B", "C"], .....: index=pd.date_range("1/1/2000", periods=10), .....: ) .....: In [186]: tsdf.iloc[3:7] = np.nan In [187]: tsdf Out[187]: A B C 2000-01-01 -0.428759 -0.864890 -0.675341 2000-01-02 -0.168731 1.338144 -1.279321 2000-01-03 -1.621034 0.438107 0.903794 2000-01-04 NaN NaN NaN 2000-01-05 NaN NaN NaN 2000-01-06 NaN NaN NaN 2000-01-07 NaN NaN NaN 2000-01-08 0.254374 -1.240447 -0.201052 2000-01-09 -0.157795 0.791197 -1.144209 2000-01-10 -0.030876 0.371900 0.061932

Transform the entire frame. .transform() allows input functions as: a NumPy function, a string function name or a user defined function.

In [188]: tsdf.transform(np.abs) Out[188]: A B C 2000-01-01 0.428759 0.864890 0.675341 2000-01-02 0.168731 1.338144 1.279321 2000-01-03 1.621034 0.438107 0.903794 2000-01-04 NaN NaN NaN 2000-01-05 NaN NaN NaN 2000-01-06 NaN NaN NaN 2000-01-07 NaN NaN NaN 2000-01-08 0.254374 1.240447 0.201052 2000-01-09 0.157795 0.791197 1.144209 2000-01-10 0.030876 0.371900 0.061932 In [189]: tsdf.transform("abs") Out[189]: A B C 2000-01-01 0.428759 0.864890 0.675341 2000-01-02 0.168731 1.338144 1.279321 2000-01-03 1.621034 0.438107 0.903794 2000-01-04 NaN NaN NaN 2000-01-05 NaN NaN NaN 2000-01-06 NaN NaN NaN 2000-01-07 NaN NaN NaN 2000-01-08 0.254374 1.240447 0.201052 2000-01-09 0.157795 0.791197 1.144209 2000-01-10 0.030876 0.371900 0.061932 In [190]: tsdf.transform(lambda x: x.abs()) Out[190]: A B C 2000-01-01 0.428759 0.864890 0.675341 2000-01-02 0.168731 1.338144 1.279321 2000-01-03 1.621034 0.438107 0.903794 2000-01-04 NaN NaN NaN 2000-01-05 NaN NaN NaN 2000-01-06 NaN NaN NaN 2000-01-07 NaN NaN NaN 2000-01-08 0.254374 1.240447 0.201052 2000-01-09 0.157795 0.791197 1.144209 2000-01-10 0.030876 0.371900 0.061932

Here transform() received a single function; this is equivalent to a ufunc application.

In [191]: np.abs(tsdf) Out[191]: A B C 2000-01-01 0.428759 0.864890 0.675341 2000-01-02 0.168731 1.338144 1.279321 2000-01-03 1.621034 0.438107 0.903794 2000-01-04 NaN NaN NaN 2000-01-05 NaN NaN NaN 2000-01-06 NaN NaN NaN 2000-01-07 NaN NaN NaN 2000-01-08 0.254374 1.240447 0.201052 2000-01-09 0.157795 0.791197 1.144209 2000-01-10 0.030876 0.371900 0.061932

Passing a single function to .transform() with a Series will yield a single Series in return.

In [192]: tsdf["A"].transform(np.abs) Out[192]: 2000-01-01 0.428759 2000-01-02 0.168731 2000-01-03 1.621034 2000-01-04 NaN 2000-01-05 NaN 2000-01-06 NaN 2000-01-07 NaN 2000-01-08 0.254374 2000-01-09 0.157795 2000-01-10 0.030876 Freq: D, Name: A, dtype: float64

Applying elementwise functions#

Since not all functions can be vectorized (accept NumPy arrays and return another array or value), the methods applymap() on DataFrame and analogously map() on Series accept any Python function taking a single value and returning a single value. For example:

In [197]: df4 Out[197]: one two three a 1.394981 1.772517 NaN b 0.343054 1.912123 -0.050390 c 0.695246 1.478369 1.227435 d NaN 0.279344 -0.613172 In [198]: def f(x): .....: return len(str(x)) .....: In [199]: df4["one"].map(f) Out[199]: a 18 b 19 c 18 d 3 Name: one, dtype: int64 In [200]: df4.applymap(f) Out[200]: one two three a 18 17 3 b 19 18 20 c 18 18 16 d 3 19 19

Series.map() has an additional feature; it can be used to easily “link” or “map” values defined by a secondary series. This is closely related to merging/joining functionality:

In [201]: s = pd.Series( .....: ["six", "seven", "six", "seven", "six"], index=["a", "b", "c", "d", "e"] .....: ) .....: In [202]: t = pd.Series({"six": 6.0, "seven": 7.0}) In [203]: s Out[203]: a six b seven c six d seven e six dtype: object In [204]: s.map(t) Out[204]: a 6.0 b 7.0 c 6.0 d 7.0 e 6.0 dtype: float64

Reindexing to align with another object#

You may wish to take an object and reindex its axes to be labeled the same as another object. While the syntax for this is straightforward albeit verbose, it is a common enough operation that the reindex_like() method is available to make this simpler:

In [216]: df2 Out[216]: one two a 1.394981 1.772517 b 0.343054 1.912123 c 0.695246 1.478369 In [217]: df3 Out[217]: one two a 0.583888 0.051514 b -0.468040 0.191120 c -0.115848 -0.242634 In [218]: df.reindex_like(df2) Out[218]: one two a 1.394981 1.772517 b 0.343054 1.912123 c 0.695246 1.478369

Aligning objects with each other with align#

The align() method is the fastest way to simultaneously align two objects. It supports a join argument (related to joining and merging):

• join='outer': take the union of the indexes (default)

• join='left': use the calling object’s index

• join='right': use the passed object’s index

• join='inner': intersect the indexes

It returns a tuple with both of the reindexed Series:

In [219]: s = pd.Series(np.random.randn(5), index=["a", "b", "c", "d", "e"]) In [220]: s1 = s[:4] In [221]: s2 = s[1:] In [222]: s1.align(s2) Out[222]: (a -0.186646 b -1.692424 c -0.303893 d -1.425662 e NaN dtype: float64, a NaN b -1.692424 c -0.303893 d -1.425662 e 1.114285 dtype: float64) In [223]: s1.align(s2, join="inner") Out[223]: (b -1.692424 c -0.303893 d -1.425662 dtype: float64, b -1.692424 c -0.303893 d -1.425662 dtype: float64) In [224]: s1.align(s2, join="left") Out[224]: (a -0.186646 b -1.692424 c -0.303893 d -1.425662 dtype: float64, a NaN b -1.692424 c -0.303893 d -1.425662 dtype: float64)

For DataFrames, the join method will be applied to both the index and the columns by default:

In [225]: df.align(df2, join="inner") Out[225]: ( one two a 1.394981 1.772517 b 0.343054 1.912123 c 0.695246 1.478369, one two a 1.394981 1.772517 b 0.343054 1.912123 c 0.695246 1.478369)

You can also pass an axis option to only align on the specified axis:

In [226]: df.align(df2, join="inner", axis=0) Out[226]: ( one two three a 1.394981 1.772517 NaN b 0.343054 1.912123 -0.050390 c 0.695246 1.478369 1.227435, one two a 1.394981 1.772517 b 0.343054 1.912123 c 0.695246 1.478369)

If you pass a Series to DataFrame.align(), you can choose to align both objects either on the DataFrame’s index or columns using the axis argument:

In [227]: df.align(df2.iloc[0], axis=1) Out[227]: ( one three two a 1.394981 NaN 1.772517 b 0.343054 -0.050390 1.912123 c 0.695246 1.227435 1.478369 d NaN -0.613172 0.279344, one 1.394981 three NaN two 1.772517 Name: a, dtype: float64)

Filling while reindexing#

reindex() takes an optional parameter method which is a filling method chosen from the following table:

Method

Action

We illustrate these fill methods on a simple Series:

In [228]: rng = pd.date_range("1/3/2000", periods=8) In [229]: ts = pd.Series(np.random.randn(8), index=rng) In [230]: ts2 = ts[[0, 3, 6]] In [231]: ts Out[231]: 2000-01-03 0.183051 2000-01-04 0.400528 2000-01-05 -0.015083 2000-01-06 2.395489 2000-01-07 1.414806 2000-01-08 0.118428 2000-01-09 0.733639 2000-01-10 -0.936077 Freq: D, dtype: float64 In [232]: ts2 Out[232]: 2000-01-03 0.183051 2000-01-06 2.395489 2000-01-09 0.733639 Freq: 3D, dtype: float64 In [233]: ts2.reindex(ts.index) Out[233]: 2000-01-03 0.183051 2000-01-04 NaN 2000-01-05 NaN 2000-01-06 2.395489 2000-01-07 NaN 2000-01-08 NaN 2000-01-09 0.733639 2000-01-10 NaN Freq: D, dtype: float64 In [234]: ts2.reindex(ts.index, method="ffill") Out[234]: 2000-01-03 0.183051 2000-01-04 0.183051 2000-01-05 0.183051 2000-01-06 2.395489 2000-01-07 2.395489 2000-01-08 2.395489 2000-01-09 0.733639 2000-01-10 0.733639 Freq: D, dtype: float64 In [235]: ts2.reindex(ts.index, method="bfill") Out[235]: 2000-01-03 0.183051 2000-01-04 2.395489 2000-01-05 2.395489 2000-01-06 2.395489 2000-01-07 0.733639 2000-01-08 0.733639 2000-01-09 0.733639 2000-01-10 NaN Freq: D, dtype: float64 In [236]: ts2.reindex(ts.index, method="nearest") Out[236]: 2000-01-03 0.183051 2000-01-04 0.183051 2000-01-05 2.395489 2000-01-06 2.395489 2000-01-07 2.395489 2000-01-08 0.733639 2000-01-09 0.733639 2000-01-10 0.733639 Freq: D, dtype: float64

These methods require that the indexes are ordered increasing or decreasing.

Note that the same result could have been achieved using fillna (except for method='nearest') or interpolate:

In [237]: ts2.reindex(ts.index).fillna(method="ffill") Out[237]: 2000-01-03 0.183051 2000-01-04 0.183051 2000-01-05 0.183051 2000-01-06 2.395489 2000-01-07 2.395489 2000-01-08 2.395489 2000-01-09 0.733639 2000-01-10 0.733639 Freq: D, dtype: float64

reindex() will raise a ValueError if the index is not monotonically increasing or decreasing. fillna() and interpolate() will not perform any checks on the order of the index.

Limits on filling while reindexing#

The limit and tolerance arguments provide additional control over filling while reindexing. Limit specifies the maximum count of consecutive matches:

In [238]: ts2.reindex(ts.index, method="ffill", limit=1) Out[238]: 2000-01-03 0.183051 2000-01-04 0.183051 2000-01-05 NaN 2000-01-06 2.395489 2000-01-07 2.395489 2000-01-08 NaN 2000-01-09 0.733639 2000-01-10 0.733639 Freq: D, dtype: float64

In contrast, tolerance specifies the maximum distance between the index and indexer values:

In [239]: ts2.reindex(ts.index, method="ffill", tolerance="1 day") Out[239]: 2000-01-03 0.183051 2000-01-04 0.183051 2000-01-05 NaN 2000-01-06 2.395489 2000-01-07 2.395489 2000-01-08 NaN 2000-01-09 0.733639 2000-01-10 0.733639 Freq: D, dtype: float64

Notice that when used on a DatetimeIndex, TimedeltaIndex or PeriodIndex, tolerance will coerced into a Timedelta if possible. This allows you to specify tolerance with appropriate strings.

Dropping labels from an axis#

A method closely related to reindex is the drop() function. It removes a set of labels from an axis:

In [240]: df Out[240]: one two three a 1.394981 1.772517 NaN b 0.343054 1.912123 -0.050390 c 0.695246 1.478369 1.227435 d NaN 0.279344 -0.613172 In [241]: df.drop(["a", "d"], axis=0) Out[241]: one two three b 0.343054 1.912123 -0.050390 c 0.695246 1.478369 1.227435 In [242]: df.drop(["one"], axis=1) Out[242]: two three a 1.772517 NaN b 1.912123 -0.050390 c 1.478369 1.227435 d 0.279344 -0.613172

Note that the following also works, but is a bit less obvious / clean:

In [243]: df.reindex(df.index.difference(["a", "d"])) Out[243]: one two three b 0.343054 1.912123 -0.050390 c 0.695246 1.478369 1.227435

Renaming / mapping labels#

The rename() method allows you to relabel an axis based on some mapping (a dict or Series) or an arbitrary function.

In [244]: s Out[244]: a -0.186646 b -1.692424 c -0.303893 d -1.425662 e 1.114285 dtype: float64 In [245]: s.rename(str.upper) Out[245]: A -0.186646 B -1.692424 C -0.303893 D -1.425662 E 1.114285 dtype: float64

If you pass a function, it must return a value when called with any of the labels (and must produce a set of unique values). A dict or Series can also be used:

In [246]: df.rename( .....: columns={"one": "foo", "two": "bar"}, .....: index={"a": "apple", "b": "banana", "d": "durian"}, .....: ) .....: Out[246]: foo bar three apple 1.394981 1.772517 NaN banana 0.343054 1.912123 -0.050390 c 0.695246 1.478369 1.227435 durian NaN 0.279344 -0.613172

If the mapping doesn’t include a column/index label, it isn’t renamed. Note that extra labels in the mapping don’t throw an error.

DataFrame.rename() also supports an “axis-style” calling convention, where you specify a single mapper and the axis to apply that mapping to.

In [247]: df.rename({"one": "foo", "two": "bar"}, axis="columns") Out[247]: foo bar three a 1.394981 1.772517 NaN b 0.343054 1.912123 -0.050390 c 0.695246 1.478369 1.227435 d NaN 0.279344 -0.613172 In [248]: df.rename({"a": "apple", "b": "banana", "d": "durian"}, axis="index") Out[248]: one two three apple 1.394981 1.772517 NaN banana 0.343054 1.912123 -0.050390 c 0.695246 1.478369 1.227435 durian NaN 0.279344 -0.613172

The rename() method also provides an inplace named parameter that is by default False and copies the underlying data. Pass inplace=True to rename the data in place.

Finally, rename() also accepts a scalar or list-like for altering the Series.name attribute.

In [249]: s.rename("scalar-name") Out[249]: a -0.186646 b -1.692424 c -0.303893 d -1.425662 e 1.114285 Name: scalar-name, dtype: float64

The methods DataFrame.rename_axis() and Series.rename_axis() allow specific names of a MultiIndex to be changed (as opposed to the labels).

In [250]: df = pd.DataFrame( .....: {"x": [1, 2, 3, 4, 5, 6], "y": [10, 20, 30, 40, 50, 60]}, .....: index=pd.MultiIndex.from_product( .....: [["a", "b", "c"], [1, 2]], names=["let", "num"] .....: ), .....: ) .....: In [251]: df Out[251]: x y let num a 1 1 10 2 2 20 b 1 3 30 2 4 40 c 1 5 50 2 6 60 In [252]: df.rename_axis(index={"let": "abc"}) Out[252]: x y abc num a 1 1 10 2 2 20 b 1 3 30 2 4 40 c 1 5 50 2 6 60 In [253]: df.rename_axis(index=str.upper) Out[253]: x y LET NUM a 1 1 10 2 2 20 b 1 3 30 2 4 40 c 1 5 50 2 6 60

items#

Consistent with the dict-like interface, items() iterates through key-value pairs:

• Series: (index, scalar value) pairs

• DataFrame: (column, Series) pairs

For example:

In [259]: for label, ser in df.items(): .....: print(label) .....: print(ser) .....: a 0 1 1 2 2 3 Name: a, dtype: int64 b 0 a 1 b 2 c Name: b, dtype: object

iterrows#

iterrows() allows you to iterate through the rows of a DataFrame as Series objects. It returns an iterator yielding each index value along with a Series containing the data in each row:

In [260]: for row_index, row in df.iterrows(): .....: print(row_index, row, sep="\n") .....: 0 a 1 b a Name: 0, dtype: object 1 a 2 b b Name: 1, dtype: object 2 a 3 b c Name: 2, dtype: object

Note

Because iterrows() returns a Series for each row, it does not preserve dtypes across the rows (dtypes are preserved across columns for DataFrames). For example,

In [261]: df_orig = pd.DataFrame([[1, 1.5]], columns=["int", "float"]) In [262]: df_orig.dtypes Out[262]: int int64 float float64 dtype: object In [263]: row = next(df_orig.iterrows())[1] In [264]: row Out[264]: int 1.0 float 1.5 Name: 0, dtype: float64

All values in row, returned as a Series, are now upcasted to floats, also the original integer value in column x:

In [265]: row["int"].dtype Out[265]: dtype('float64') In [266]: df_orig["int"].dtype Out[266]: dtype('int64')

To preserve dtypes while iterating over the rows, it is better to use itertuples() which returns namedtuples of the values and which is generally much faster than iterrows().

For instance, a contrived way to transpose the DataFrame would be:

In [267]: df2 = pd.DataFrame({"x": [1, 2, 3], "y": [4, 5, 6]}) In [268]: print(df2) x y 0 1 4 1 2 5 2 3 6 In [269]: print(df2.T) 0 1 2 x 1 2 3 y 4 5 6 In [270]: df2_t = pd.DataFrame({idx: values for idx, values in df2.iterrows()}) In [271]: print(df2_t) 0 1 2 x 1 2 3 y 4 5 6

itertuples#

The itertuples() method will return an iterator yielding a namedtuple for each row in the DataFrame. The first element of the tuple will be the row’s corresponding index value, while the remaining values are the row values.

For instance:

In [272]: for row in df.itertuples(): .....: print(row) .....: Pandas(Index=0, a=1, b='a') Pandas(Index=1, a=2, b='b') Pandas(Index=2, a=3, b='c')

This method does not convert the row to a Series object; it merely returns the values inside a namedtuple. Therefore, itertuples() preserves the data type of the values and is generally faster as iterrows().

Note

The column names will be renamed to positional names if they are invalid Python identifiers, repeated, or start with an underscore. With a large number of columns (>255), regular tuples are returned.

By index#

The Series.sort_index() and DataFrame.sort_index() methods are used to sort a pandas object by its index levels.

In [300]: df = pd.DataFrame( .....: { .....: "one": pd.Series(np.random.randn(3), index=["a", "b", "c"]), .....: "two": pd.Series(np.random.randn(4), index=["a", "b", "c", "d"]), .....: "three": pd.Series(np.random.randn(3), index=["b", "c", "d"]), .....: } .....: ) .....: In [301]: unsorted_df = df.reindex( .....: index=["a", "d", "c", "b"], columns=["three", "two", "one"] .....: ) .....: In [302]: unsorted_df Out[302]: three two one a NaN -1.152244 0.562973 d -0.252916 -0.109597 NaN c 1.273388 -0.167123 0.640382 b -0.098217 0.009797 -1.299504 # DataFrame In [303]: unsorted_df.sort_index() Out[303]: three two one a NaN -1.152244 0.562973 b -0.098217 0.009797 -1.299504 c 1.273388 -0.167123 0.640382 d -0.252916 -0.109597 NaN In [304]: unsorted_df.sort_index(ascending=False) Out[304]: three two one d -0.252916 -0.109597 NaN c 1.273388 -0.167123 0.640382 b -0.098217 0.009797 -1.299504 a NaN -1.152244 0.562973 In [305]: unsorted_df.sort_index(axis=1) Out[305]: one three two a 0.562973 NaN -1.152244 d NaN -0.252916 -0.109597 c 0.640382 1.273388 -0.167123 b -1.299504 -0.098217 0.009797 # Series In [306]: unsorted_df["three"].sort_index() Out[306]: a NaN b -0.098217 c 1.273388 d -0.252916 Name: three, dtype: float64

New in version 1.1.0.

Sorting by index also supports a key parameter that takes a callable function to apply to the index being sorted. For MultiIndex objects, the key is applied per-level to the levels specified by level.

In [307]: s1 = pd.DataFrame({"a": ["B", "a", "C"], "b": [1, 2, 3], "c": [2, 3, 4]}).set_index( .....: list("ab") .....: ) .....: In [308]: s1 Out[308]: c a b B 1 2 a 2 3 C 3 4

In [309]: s1.sort_index(level="a") Out[309]: c a b B 1 2 C 3 4 a 2 3 In [310]: s1.sort_index(level="a", key=lambda idx: idx.str.lower()) Out[310]: c a b a 2 3 B 1 2 C 3 4

For information on key sorting by value, see value sorting.

By values#

The Series.sort_values() method is used to sort a Series by its values. The DataFrame.sort_values() method is used to sort a DataFrame by its column or row values. The optional by parameter to DataFrame.sort_values() may used to specify one or more columns to use to determine the sorted order.

In [311]: df1 = pd.DataFrame( .....: {"one": [2, 1, 1, 1], "two": [1, 3, 2, 4], "three": [5, 4, 3, 2]} .....: ) .....: In [312]: df1.sort_values(by="two") Out[312]: one two three 0 2 1 5 2 1 2 3 1 1 3 4 3 1 4 2

The by parameter can take a list of column names, e.g.:

In [313]: df1[["one", "two", "three"]].sort_values(by=["one", "two"]) Out[313]: one two three 2 1 2 3 1 1 3 4 3 1 4 2 0 2 1 5

These methods have special treatment of NA values via the na_position argument:

In [314]: s[2] = np.nan In [315]: s.sort_values() Out[315]: 0 A 3 Aaba 1 B 4 Baca 6 CABA 8 cat 7 dog 2 <NA> 5 <NA> dtype: string In [316]: s.sort_values(na_position="first") Out[316]: 2 <NA> 5 <NA> 0 A 3 Aaba 1 B 4 Baca 6 CABA 8 cat 7 dog dtype: string

New in version 1.1.0.

Sorting also supports a key parameter that takes a callable function to apply to the values being sorted.

In [317]: s1 = pd.Series(["B", "a", "C"])

In [318]: s1.sort_values() Out[318]: 0 B 2 C 1 a dtype: object In [319]: s1.sort_values(key=lambda x: x.str.lower()) Out[319]: 1 a 0 B 2 C dtype: object

key will be given the Series of values and should return a Series or array of the same shape with the transformed values. For DataFrame objects, the key is applied per column, so the key should still expect a Series and return a Series, e.g.

In [320]: df = pd.DataFrame({"a": ["B", "a", "C"], "b": [1, 2, 3]})

In [321]: df.sort_values(by="a") Out[321]: a b 0 B 1 2 C 3 1 a 2 In [322]: df.sort_values(by="a", key=lambda col: col.str.lower()) Out[322]: a b 1 a 2 0 B 1 2 C 3

The name or type of each column can be used to apply different functions to different columns.

By indexes and values#

Strings passed as the by parameter to DataFrame.sort_values() may refer to either columns or index level names.

# Build MultiIndex In [323]: idx = pd.MultiIndex.from_tuples( .....: [("a", 1), ("a", 2), ("a", 2), ("b", 2), ("b", 1), ("b", 1)] .....: ) .....: In [324]: idx.names = ["first", "second"] # Build DataFrame In [325]: df_multi = pd.DataFrame({"A": np.arange(6, 0, -1)}, index=idx) In [326]: df_multi Out[326]: A first second a 1 6 2 5 2 4 b 2 3 1 2 1 1

Sort by ‘second’ (index) and ‘A’ (column)

In [327]: df_multi.sort_values(by=["second", "A"]) Out[327]: A first second b 1 1 1 2 a 1 6 b 2 3 a 2 4 2 5

Note

If a string matches both a column name and an index level name then a warning is issued and the column takes precedence. This will result in an ambiguity error in a future version.

searchsorted#

Series has the searchsorted() method, which works similarly to numpy.ndarray.searchsorted().

In [328]: ser = pd.Series([1, 2, 3]) In [329]: ser.searchsorted([0, 3]) Out[329]: array([0, 2]) In [330]: ser.searchsorted([0, 4]) Out[330]: array([0, 3]) In [331]: ser.searchsorted([1, 3], side="right") Out[331]: array([1, 3]) In [332]: ser.searchsorted([1, 3], side="left") Out[332]: array([0, 2]) In [333]: ser = pd.Series([3, 1, 2]) In [334]: ser.searchsorted([0, 3], sorter=np.argsort(ser)) Out[334]: array([0, 2])

smallest / largest values#

Series has the nsmallest() and nlargest() methods which return the smallest or largest \(n\) values. For a large Series this can be much faster than sorting the entire Series and calling head(n) on the result.

In [335]: s = pd.Series(np.random.permutation(10)) In [336]: s Out[336]: 0 2 1 0 2 3 3 7 4 1 5 5 6 9 7 6 8 8 9 4 dtype: int64 In [337]: s.sort_values() Out[337]: 1 0 4 1 0 2 2 3 9 4 5 5 7 6 3 7 8 8 6 9 dtype: int64 In [338]: s.nsmallest(3) Out[338]: 1 0 4 1 0 2 dtype: int64 In [339]: s.nlargest(3) Out[339]: 6 9 8 8 3 7 dtype: int64

DataFrame also has the nlargest and nsmallest methods.

In [340]: df = pd.DataFrame( .....: { .....: "a": [-2, -1, 1, 10, 8, 11, -1], .....: "b": list("abdceff"), .....: "c": [1.0, 2.0, 4.0, 3.2, np.nan, 3.0, 4.0], .....: } .....: ) .....: In [341]: df.nlargest(3, "a") Out[341]: a b c 5 11 f 3.0 3 10 c 3.2 4 8 e NaN In [342]: df.nlargest(5, ["a", "c"]) Out[342]: a b c 5 11 f 3.0 3 10 c 3.2 4 8 e NaN 2 1 d 4.0 6 -1 f 4.0 In [343]: df.nsmallest(3, "a") Out[343]: a b c 0 -2 a 1.0 1 -1 b 2.0 6 -1 f 4.0 In [344]: df.nsmallest(5, ["a", "c"]) Out[344]: a b c 0 -2 a 1.0 1 -1 b 2.0 6 -1 f 4.0 2 1 d 4.0 4 8 e NaN

Sorting by a MultiIndex column#

You must be explicit about sorting when the column is a MultiIndex, and fully specify all levels to by.

In [345]: df1.columns = pd.MultiIndex.from_tuples( .....: [("a", "one"), ("a", "two"), ("b", "three")] .....: ) .....: In [346]: df1.sort_values(by=("a", "two")) Out[346]: a b one two three 0 2 1 5 2 1 2 3 1 1 3 4 3 1 4 2