import warnings
warnings.filterwarnings('ignore')
import pandas as pd
import numpy as np
import openml

# Load a primary data example: synthetic Net Promoter Score (NPS) survey
np.random.seed(42)
nps_df = pd.DataFrame({
    'CustomerID': range(1, 501),
    'Age': np.random.randint(18, 70, 500),
    'Region': np.random.choice(['North', 'South', 'East', 'West'], 500),
    'NPS_Score': np.random.randint(0, 11, 500)
})
print(nps_df.head(3))

# Load a secondary data example: OpenML customer satisfaction survey data
dataset = openml.datasets.get_dataset(42178)
csat_df, _, _, _ = dataset.get_data(dataset_format='dataframe')
print(csat_df.shape)
print(csat_df.head(3))

# Load a behavioral/secondary data example: online retail transactions from UCI
url = 'https://archive.ics.uci.edu/ml/machine-learning-databases/00502/online_retail_II.xlsx'
retail_df = pd.read_excel(url, sheet_name='Year 2010-2011')
retail_df['InvoiceDate'] = pd.to_datetime(retail_df['InvoiceDate'])
print(retail_df.head(3))

avg_nps = nps_df['NPS_Score'].mean()
print(f'Overall mean NPS score: {avg_nps:.2f}')

satisfied_prop = (csat_df['Churn'] == 'No').mean()
print(f'Proportion of customers who did not churn: {satisfied_prop:.2%}')

region_nps = nps_df.groupby('Region')['NPS_Score'].mean()
print(region_nps)

import matplotlib.pyplot as plt
plt.hist(nps_df['NPS_Score'], bins=11, edgecolor='k')
plt.xlabel('NPS Score')
plt.ylabel('Frequency')
plt.title('Distribution of NPS Survey Scores')
plt.show()

churn_by_contract = csat_df.groupby('Contract')['Churn'].value_counts(normalize=True).unstack()['Yes']
print(churn_by_contract)

retail_df_nonnull = retail_df.dropna(subset=['Customer ID'])
order_values = retail_df_nonnull.groupby('Customer ID').apply(lambda g: (g['Quantity'] * g['Price']).sum())
avg_order_value = order_values.mean()
print(f'Average order size per customer: GBP {avg_order_value:.2f}')

def nps_segment(score):
    if score >= 9: return 'Promoter'
    elif score >= 7: return 'Passive'
    else: return 'Detractor'
nps_df['Segment'] = nps_df['NPS_Score'].apply(nps_segment)
print(nps_df['Segment'].value_counts())

crosstab = pd.crosstab(csat_df['PaperlessBilling'], csat_df['Churn'], margins=True, normalize='index')
print(crosstab)

# Error Handling 1: Find Missing Survey Responses
missing_nps = nps_df.isnull().sum()
print('Missing values by column:\n', missing_nps)

# Error Handling 2: Check for Incorrect Aggregation (Grouping by wrong column)
try:
    wrong_group = nps_df.groupby('NPS_Score')['Region'].count()
    print(wrong_group.head())
except Exception as e:
    print(f'Error: {e}')

# Error Handling 3: Misinterpreting Likert Scales or NPS Scores
if nps_df['NPS_Score'].min() < 0 or nps_df['NPS_Score'].max() > 10:
    print('Warning: NPS scores must be between 0 and 10 (inclusive)!')
else:
    print('NPS score range is valid.')

# Pattern: Time Trend Analysis of Order Value (secondary behavioral data)
retail_df_nonnull['Month'] = retail_df_nonnull['InvoiceDate'].dt.to_period('M')
monthly_sales = retail_df_nonnull.groupby('Month').apply(lambda g: (g['Quantity'] * g['Price']).sum())
monthly_sales.plot(kind='line', marker='o')
plt.ylabel('Total Sales (GBP)')
plt.title('Monthly Sales Trend (2010-2011)')
plt.show()

promoter_counts = nps_df[nps_df['Segment'] == 'Promoter'].groupby('Region').size()
top_region = promoter_counts.idxmax()
print(f'Region with most Promoters: {top_region}')
print('Recommendation: Focus new referral campaigns in this region to build on current loyalty.')