susanli2016

unique_customers = grouped_df.CustomerID.unique()
customer_ids = dict(zip(unique_customers, np.arange(unique_customers.shape[0], dtype=np.int32)))

unique_items = grouped_df.StockCode.unique()
item_ids = dict(zip(unique_items, np.arange(unique_items.shape[0], dtype=np.int32)))

grouped_df['customer_id'] = grouped_df.CustomerID.apply(lambda i: customer_ids[i])
grouped_df['item_id'] = grouped_df.StockCode.apply(lambda i: item_ids[i])

sparse_item_customer = sparse.csr_matrix((grouped_df['Quantity'].astype(float), (grouped_df['item_id'], grouped_df['customer_id'])))

susanli2016

print(f'Number of unique customers: {grouped_df.CustomerID.nunique()}')
print(f'Number of unique items: {grouped_df.StockCode.nunique()}')

print(f'Average purchase quantity per interaction: {int(grouped_df.Quantity.mean())}')
print(f'Minimum purchase quantity per interaction: {grouped_df.Quantity.min()}')
print(f'Maximum purchase quantity per interaction: {grouped_df.Quantity.max()}')

susanli2016

import plotly.express as px

fig = px.histogram(grouped_df, x='Quantity', title='Distribution of the purchase quantity', nbins=500)
fig.show();

susanli2016

retail_df = retail_df[retail_df['CustomerID'].notna()]
grouped_df = retail_df[['CustomerID', 'StockCode', 'Description', 'Quantity']].groupby(['CustomerID', 'StockCode', 'Description']).sum().reset_index()
grouped_df.loc[grouped_df['Quantity'] == 0, ['Quantity']] = 1
grouped_df = grouped_df.loc[grouped_df['Quantity'] > 0]

susanli2016

import sys
import pandas as pd
import numpy as np
import scipy.sparse as sparse
from scipy.sparse.linalg import spsolve
import random
from sklearn import metrics
from sklearn.preprocessing import MinMaxScaler
import implicit

susanli2016

from prince import MCA

df_cat = df_clean[['vote16', 'vote', 'confecon', 'self_ideology', 'immignum', 'health_care', 'worry_covid', 
                   'worry_covid_economy', 'partyID', 'sex', 'science', 'universal_income', 'freecol']].dropna()
mca = MCA(n_components=2)
mca = mca.fit(df_cat)

mca.column_coordinates(df_cat).sort_values(0)

susanli2016

from prince import PCA

df_ft = df_clean.loc[:, 'fttrump1': 'ftdemocraticparty'].dropna()
pca = PCA(n_components=2, 
          rescale_with_mean=True,
          rescale_with_std=True)
pca = pca.fit(df_ft)

loadings = pca.column_correlations(df_ft)
loadings.columns = ['dim1', 'dim2']

susanli2016

stats.chi2_contingency(pd.crosstab(df_clean['favor_both'], df_clean['self_ideology']).values)

susanli2016

round(pd.crosstab(df_clean['favor_both'], df_clean['self_ideology'], normalize='columns')*100,2)

susanli2016

stats.f_oneway(df_clean.query("partyID=='Democrat'").age.dropna(),
               df_clean.query("partyID=='Republican'").age.dropna(),
               df_clean.query("partyID=='independent'").age.dropna())