Implement (DataFrame|Series).plot.hist in plotly

databricks/koalas/plot/core.py

@@ -18,6 +18,8 @@
 
 import pandas as pd
 import numpy as np
+
+from databricks.koalas.utils import name_like_string
 from pyspark.ml.feature import Bucketizer
 from pyspark.sql import functions as F
 from pandas.core.base import PandasObject
@@ -143,8 +145,11 @@ def compute_hist(kdf, bins):
 
         sdf = kdf._internal.spark_frame
         scols = []
+        input_column_names = []
         for label in kdf._internal.column_labels:
-            scols.append(kdf._internal.spark_column_for(label))
+            input_column_name = name_like_string(label)
+            input_column_names.append(input_column_name)
+            scols.append(kdf._internal.spark_column_for(label).alias(input_column_name))
         sdf = sdf.select(*scols)
 
         # 1. Make the bucket output flat to:
@@ -243,7 +248,7 @@ def compute_hist(kdf, bins):
         #     |0                |
         #     +-----------------+
         output_series = []
-        for i, bucket_name in enumerate(bucket_names):
+        for i, (input_column_name, bucket_name) in enumerate(zip(input_column_names, bucket_names)):
             current_bucket_result = result[result["__group_id"] == i]
             # generates a pandas DF with one row for each bin
             # we need this as some of the bins may be empty
@@ -252,8 +257,8 @@ def compute_hist(kdf, bins):
             pdf = indexes.merge(current_bucket_result, how="left", on=["__bucket"]).fillna(0)[
                 ["count"]
             ]
-            pdf.columns = [bucket_name]
-            output_series.append(pdf[bucket_name])
+            pdf.columns = [input_column_name]
+            output_series.append(pdf[input_column_name])
 
         return output_series
 
@@ -363,6 +368,8 @@ def __call__(self, kind="line", backend=None, **kwargs):
                 "scatter": TopNPlotBase().get_top_n,
                 "area": SampledPlotBase().get_sampled,
                 "line": SampledPlotBase().get_sampled,
+                # if histogram is not supported, the backend will throw an exception
+                "hist": lambda data: data,
             }
             if not data_preprocessor_map[kind]:
                 raise NotImplementedError(

databricks/koalas/plot/plotly.py

@@ -15,12 +15,17 @@
 #
 import pandas as pd
 
+from databricks.koalas.plot import HistogramPlotBase
+
 
 def plot_plotly(origin_plot):
     def plot(data, kind, **kwargs):
         # Koalas specific plots
         if kind == "pie":
             return plot_pie(data, **kwargs)
+        if kind == "hist":
+            # Note that here data is a Koalas DataFrame or Series unlike other type of plots.
+            return plot_histogram(data, **kwargs)
 
         # Other plots.
         return origin_plot(data, kind, **kwargs)
@@ -49,3 +54,50 @@ def plot_pie(data, **kwargs):
         )
     else:
         raise RuntimeError("Unexpected type: [%s]" % type(data))
+
+
+def plot_histogram(data, **kwargs):
+    from plotly import express
+    from plotly.subplots import make_subplots
+    import plotly.graph_objs as go
+    from databricks import koalas as ks
+
+    assert "bins" in kwargs
+    bins = kwargs["bins"]
+    data, bins = HistogramPlotBase.prepare_hist_data(data, bins)
+
+    is_single_column = False
+    if isinstance(data, ks.Series):
+        is_single_column = True
+        kdf = data.to_frame()
+    elif isinstance(data, ks.DataFrame):
+        kdf = data
+    else:
+        raise RuntimeError("Unexpected type: [%s]" % type(data))
+
+    output_series = HistogramPlotBase.compute_hist(kdf, bins)
+    bins = 0.5 * (bins[:-1] + bins[1:])
+    if is_single_column:
+        output_series = list(output_series)
+        assert len(output_series) == 1
+        output_series = output_series[0]
+        return express.bar(
+            x=bins, y=output_series, labels={"x": str(output_series.name), "y": "count"}
+        )
+    else:
+        output_series = list(output_series)
+        fig = make_subplots(rows=1, cols=len(output_series))
+
+        for i, series in enumerate(output_series):
+            fig.add_trace(go.Bar(x=bins, y=series,), row=1, col=i + 1)
+
+        for i, series in enumerate(output_series):
+            if i == 0:
+                xaxis = "xaxis"
+                yaxis = "yaxis"
+            else:
+                xaxis = "xaxis%s" % (i + 1)
+                yaxis = "yaxis%s" % (i + 1)
+            fig["layout"][xaxis]["title"] = str(series.name)
+            fig["layout"][yaxis]["title"] = "count"
+        return fig