IIS/assets/eda/eda.ipynb

import pandas as pd
import matplotlib as plt
import seaborn as sns
import numpy as np

Загрузка и знакомство с данными

# dataset https://www.kaggle.com/datasets/mrdaniilak/russia-real-estate-20182021/data 

df = pd.read_csv('data/all_v2.csv')

df.head(10)

	price	date	time	geo_lat	geo_lon	region	building_type	level	levels	rooms	area	kitchen_area	object_type
0	6050000	2018-02-19	20:00:21	59.805808	30.376141	2661	1	8	10	3	82.6	10.8	1
1	8650000	2018-02-27	12:04:54	55.683807	37.297405	81	3	5	24	2	69.1	12.0	1
2	4000000	2018-02-28	15:44:00	56.295250	44.061637	2871	1	5	9	3	66.0	10.0	1
3	1850000	2018-03-01	11:24:52	44.996132	39.074783	2843	4	12	16	2	38.0	5.0	11
4	5450000	2018-03-01	17:42:43	55.918767	37.984642	81	3	13	14	2	60.0	10.0	1
5	3300000	2018-03-02	21:18:42	55.908253	37.726448	81	1	4	5	1	32.0	6.0	1
6	4704280	2018-03-04	12:35:25	55.621097	37.431002	3	2	1	25	1	31.7	6.0	11
7	3600000	2018-03-04	20:52:38	59.875526	30.395457	2661	1	2	5	1	31.1	6.0	1
8	3390000	2018-03-05	07:07:05	53.195031	50.106952	3106	2	4	24	2	64.0	13.0	11
9	2800000	2018-03-06	09:57:10	55.736972	38.846457	81	1	9	10	2	55.0	8.0	1

Очистка данных

Анализ признаков для модели

https://seaborn.pydata.org/examples/index.html - галерея примеров

histplot

heatmap

Групповые операции

def flat_index(df_stats): 
    df_stats.columns = df_stats.columns.get_level_values(0) + '_' +  df_stats.columns.get_level_values(1) 
    df_stats.columns = df_stats.columns.to_flat_index() 
    df_stats.reset_index(inplace=True) 
    return df_stats

lineplot

subplots

fig, axs = plt.pyplot.subplots(2,2)
fig.tight_layout(pad=1)
fig.set_size_inches(16.5, 14, forward=True)

displot

for col in categorial_cols:
    print(f'Unique categories in {col}: {df[col].nunique()}')dd

histplot

Bokeh

https://bokeh.org/

from bokeh.plotting import figure, show
from bokeh.models import ColumnDataSource,  HoverTool, Legend
from bokeh.io import output_notebook 
output_notebook()

Выводы после EDA