변수가 많아 하나하나씩 code를 작성하여 feature를 확인하기 어려워 streamlit을 활용해봤습니다.(간단한 histogram과 line plot 기능만 있으니 다른 plot이 필요한 분들은 custom하셔서 사용하세요)

streamlit을 활용하기 때문에 해당 library를 설치 이후에 사용하세요

(pip install streamlit 으로 간단하게 설치 가능합니다.)

app.py source code

import streamlit as st
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns


st.title("Data viewer")


train_df = pd.read_csv("../data/train.csv")
test_df = pd.read_csv("../data/test.csv")


x_feature_info_df = pd.read_csv("../data/meta/x_feature_info.csv")
y_feature_info_df = pd.read_csv("../data/meta/y_feature_info.csv")


x_feature_info_dict = x_feature_info_df.set_index("Feature").to_dict()["설명"]
y_feature_info_dict = y_feature_info_df.set_index("Feature").to_dict()["설명"]



X_feature_name = train_df.columns[1:-14]
y_feature_name = train_df.columns[-14:]



selected_x_feature_name = st.multiselect("Select feature name", X_feature_name)
selected_y_label_name = st.multiselect("Select label name", y_feature_name)



def feature_describe(name_list: list, info_dict: dict, option: str):
    for name in name_list:
        st.write(f"{name} (" + info_dict[name] + ")")
        if option != "null":
            feature_plot(name, option)



def feature_plot(feature_name: str, option: str):
    st.subheader("Train")
    fig, ax = plt.subplots(figsize=(30, 5))
    if option == "dist":
        sns.histplot(train_df[feature_name], label=feature_name, ax=ax, kde=True)
    else:
        ax.plot(train_df[feature_name], label=feature_name)


    ax.legend()
    st.pyplot(fig)


    if feature_name[0] != "Y":
        st.subheader("Test")
        fig, ax = plt.subplots(figsize=(30, 5))
        if option == "dist":
            sns.histplot(test_df[feature_name], label=feature_name, ax=ax, kde=True)
        else:
            ax.plot(test_df[feature_name], label=feature_name)
        ax.legend()
        st.pyplot(fig)



if len(selected_x_feature_name) > 0:
    st.subheader("X feature information")
    option = st.selectbox("Plot Type", ("null", "dist", "plot"))
    feature_describe(selected_x_feature_name, x_feature_info_dict, option)



if len(selected_y_label_name) > 0:
    st.subheader("Target feature information")
    option = st.selectbox("Plot Type", ("null", "dist", "plot"))
    feature_describe(selected_y_label_name, y_feature_info_dict, option)

경로 환경은 아래와 같습니다.

.
├── app
│   └── app.py
└── data
    ├── meta
    │   ├── x_feature_info.csv
    │   ├── y_feature_info.csv
    │   └── y_feature_spec_info.csv
    ├── open.zip
    ├── sample_submission.csv
    ├── test.csv
    └── train.csv