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

df = pd.DataFrame({
    'CustomerID':[1,2,3,4,5],
    'Feedback':[
        'Great service and friendly staff',
        'Delivery was slow and packaging was poor',
        'Excellent quality, will buy again',
        'Customer support needs improvement',
        'Good value for money']})
print(df.shape)
print(df.head(3))

df['Feedback_clean'] = df['Feedback'].str.lower().str.replace(r'[^a-z\s]', '', regex=True)
print(df[['Feedback','Feedback_clean']])

df['Feedback_wordcount'] = df['Feedback_clean'].str.split().apply(len)
print(df[['Feedback','Feedback_wordcount']])

all_words = pd.Series(' '.join(df['Feedback_clean']).split())
print(all_words.value_counts().head(10))

stopwords = ['and','was','for','the','in','is','to','of','a','an']
all_words_no_stop = all_words[~all_words.isin(stopwords)]
print(all_words_no_stop.value_counts().head(10))

import matplotlib.pyplot as plt
word_counts = all_words_no_stop.value_counts().head(5)
plt.bar(word_counts.index, word_counts.values)
plt.ylabel('Count')
plt.title('Top 5 Most Common Feedback Words')
plt.show()

theme_keywords = {
    'service':['service','support','staff','help'],
    'delivery':['delivery','packaging','slow'],
    'quality':['quality','excellent','good'],
    'value':['value','money'],
    'improvement':['improvement','needs'],
}

def tag_theme(text):
    out = []
    for theme,words in theme_keywords.items():
        if any(w in text for w in words):
            out.append(theme)
    return ', '.join(out) if out else 'other'

df['Theme'] = df['Feedback_clean'].apply(tag_theme)
print(df[['Feedback','Theme']])

theme_counts = df['Theme'].value_counts()
print(theme_counts)

theme_counts.plot(kind='bar',title='Number of Comments per Theme')
plt.ylabel('Number of Comments')
plt.show()

grouped = df.groupby('Theme').agg({'CustomerID':'count','Feedback_wordcount':'mean'})
print(grouped.rename(columns={'CustomerID':'Comment Count','Feedback_wordcount':'Avg Words'}))

nps_df = pd.DataFrame({
    'CustomerID': [1,2,3,4,5],
    'NPS_Score': [9, 3, 10, 5, 7]
})
result = df.merge(nps_df, on='CustomerID')
print(result[['Feedback','Theme','NPS_Score']])

# Bucket NPS into promoters, passives, detractors
def nps_group(x):
    if x >= 9:
        return 'Promoter'
    elif x >= 7:
        return 'Passive'
    else:
        return 'Detractor'

result['NPS_Group'] = result['NPS_Score'].apply(nps_group)
summary = result.groupby(['Theme','NPS_Group']).size().unstack().fillna(0)
print(summary)

real_df = pd.DataFrame({
    'CustomerID':[101,102,103,104,105,106,107,108],
    'Feedback':[
        'The checkout process was confusing and I could not apply my coupon',
        'Shipping was extremely fast and everything arrived in perfect condition',
        'I had trouble finding the sizing guide on mobile',
        'Would recommend. Good quality and price!',
        'My issue took too long to resolve with support',
        'The product is not as pictured',
        'Very pleasant shopping experience',
        'Got my refund right away, thank you!'
    ]
})
real_df['Feedback_clean'] = real_df['Feedback'].str.lower().str.replace(r'[^a-z\s]', '', regex=True)
print(real_df.head(3))

real_all_words = pd.Series(' '.join(real_df['Feedback_clean']).split())
real_stopwords = set(['and','the','i','in','my','to','with','was','had','is','on','for','not','as','could','a','of','it'])
words_no_stop = real_all_words[~real_all_words.isin(real_stopwords)]
print(words_no_stop.value_counts().head(10))

from collections import Counter
themes_advanced = {
    'shipping':['shipping','fast','arrived'],
    'support':['support','issue','refund','resolve','thank'],
    'website':['checkout','process','coupon','mobile','finding','guide','confusing'],
    'quality':['perfect','good','quality','pictured','product'],
    'experience':['pleasant','recommend','shopping','experience']
}

def extract_theme_adv(text):
    found = []
    for t,words in themes_advanced.items():
        if any(w in text for w in words):
            found.append(t)
    return ', '.join(found) if found else 'other'

real_df['Theme'] = real_df['Feedback_clean'].apply(extract_theme_adv)
print(real_df[['Feedback','Theme']])

theme_dist = real_df['Theme'].str.get_dummies(sep=', ').sum(axis=0).sort_values(ascending=False)
print(theme_dist)

theme_dist.plot(kind='bar',title='Theme Mentions in Customer Comments',color='steelblue')
plt.ylabel('Mentions')
plt.show()

test_df = pd.DataFrame({
    'CustomerID':[11,12,13],
    'Feedback':['Great service',np.nan,'']
})
missing = test_df['Feedback'].isnull() | (test_df['Feedback'].str.strip() == '')
print('Missing feedback entries:')
print(test_df[missing])

sample_df = pd.DataFrame({'Theme':['service','delivery','service','value','service']})
wrong_counts = sample_df.groupby('Theme').count()
print('Incorrect grouping result:')
print(wrong_counts)

right_counts = sample_df.groupby('Theme').size()
print('Correct theme counts:')
print(right_counts)

likert_scores = pd.Series(['Strongly agree', 'Agree', 'Neutral', 'Disagree', 'Strongly agree'])
score_map = {
    'Strongly agree':5,
    'Agree':4,
    'Neutral':3,
    'Disagree':2,
    'Strongly disagree':1
}
num_scores = likert_scores.map(score_map)
print(num_scores)

# Example: Cross-tabulate theme by NPS group from earlier
print(result.groupby(['Theme','NPS_Group']).size().unstack().fillna(0))

# Build a theme score (1 point for each theme per comment)
result['Theme_Score'] = result['Theme'].apply(lambda x: 0 if x=='other' else len(x.split(',')))
print(result[['Feedback','Theme','Theme_Score']])

# Example: Trend analysis (simulate monthly feedback counts by theme)
trend_df = pd.DataFrame({
    'Month':['2023-01','2023-01','2023-02','2023-02','2023-02','2023-03','2023-03'],
    'Theme':['service','delivery','service','delivery','value','value','service']
})
pivot = trend_df.pivot_table(index='Month',columns='Theme',values='Theme',aggfunc='count').fillna(0)
print(pivot)

# Workflow: From raw data to recommended business actions
final_theme_counts = real_df['Theme'].str.get_dummies(sep=', ').sum(axis=0)
main_problem = final_theme_counts.idxmax()
print('The most common feedback theme is:', main_problem)
if main_problem in ['shipping','support','website']:
    print('Business should review processes in', main_problem, 'immediately!')