Source code for prepropy.eda

import pandas as pd
import altair as alt


def eda(df, target):
    """
    Generates a dictionary to access summary statistics of the given data frame

    Parameters
    --------
    df : pandas.DataFrame
        input dataframe to be analyzed
    target : string
        target column name

    Returns
    --------
    dict
        access summary statistics of the given data frame.
    cor
        the correlation map

    Examples
    --------
    >>> from propropy import eda
    >>> url1 = "https://archive.ics.uci.edu/ml/machine-learning-databases/"
    >>> url2 = "wine-quality/winequality-red.csv"
    >>> url = url1+url2
    >>> df = pd.read_csv(url, ";")
    >>> target = "quality"
    >>> res = eda(df,quality)
    """
    # Check the dataframe input
    if not isinstance(df, pd.DataFrame):
        raise TypeError("Input data must be an instance of DataFrame")
    # Create an empty dictionary
    res = {}
    # obtain statistical information
    df_fea = df.drop(target, 1)
    num_fea = df_fea.select_dtypes("number").columns.to_list()
    cat_fea = list(set(list(df_fea.columns)) - set(num_fea))
    key_null = list(df_fea.isnull().sum().index)
    val_null = list(df_fea.isnull().sum().values)
    res["nb_missing_values"] = list(zip(key_null, val_null))
    res["nb_cat_features"] = len(cat_fea)
    res["cat_features_name"] = cat_fea
    res["nb_num_features"] = len(num_fea)
    res["num_features_name"] = num_fea
    res["nb_class"] = len(list(set(df[target])))
    class_count = df[target].value_counts(normalize=True).values
    res["class_ratio"] = list(class_count.round(4))
    # Create a pair plots with Altair
    color_lab = target + ":N"
    chart = (
        alt.Chart(df)
        .mark_circle()
        .encode(
            alt.X(alt.repeat("column"), type="quantitative"),
            alt.Y(alt.repeat("row"), type="quantitative"),
            color=color_lab,
        )
        .properties(width=100, height=100)
        .repeat(row=num_fea, column=num_fea)
    )
    res["pairplot"] = chart
    return res