Comparing Movie Ratings Data

import pandas as pd
import numpy as np

# Build the StreamVault movie ratings dataset
df = pd.DataFrame({
    "movie_id":      ["M001","M002","M003","M004","M005","M006",
                      "M007","M008","M009","M010","M011","M012"],
    "title":         ["Stellar Drift","The Last Quarter","Neon Fists",
                      "Whisper Valley","Laugh Track","Deep Meridian",
                      "City of Glass","Red Horizon","The Quiet Shore",
                      "Iron Carnival","Shadow Protocol","Sunfall"],
    "genre":         ["Sci-Fi","Drama","Action","Horror","Comedy","Drama",
                      "Sci-Fi","Action","Drama","Comedy","Horror","Sci-Fi"],
    "release_year":  [2021,2019,2022,2020,2023,2018,2022,2021,2020,2023,2019,2022],
    "runtime_min":   [132,118,105,98,94,141,127,110,124,88,103,115],
    "critic_score":  [88,91,62,79,55,94,83,58,89,48,76,85],
    "audience_score":[74,85,81,65,78,80,71,76,88,72,60,79]
})

# Basic shape and types
print("Shape:", df.shape)
print("\nData types:")
print(df.dtypes)

# Summary stats for both score columns side by side
# Selecting both score columns at once keeps the comparison clean
score_summary = df[["critic_score","audience_score"]].describe().round(1)
print("\nRating distribution summary:")
print(score_summary)

# Who scores more generously on average?
print(f"\nMean critic score:   {df['critic_score'].mean():.1f}")
print(f"Mean audience score: {df['audience_score'].mean():.1f}")
print(f"Difference:          {df['critic_score'].mean() - df['audience_score'].mean():.1f} points")

# Create a score gap column for every movie
# Positive = audiences scored higher than critics (audience-friendly film)
# Negative = critics scored higher (critic-friendly film)
df["score_gap"] = df["audience_score"] - df["critic_score"]

# Group by genre and compute mean critic score, audience score, and score gap
genre_analysis = df.groupby("genre").agg(
    mean_critic   = ("critic_score",   "mean"),
    mean_audience = ("audience_score", "mean"),
    mean_gap      = ("score_gap",      "mean"),
    film_count    = ("movie_id",       "count")
).round(1).reset_index()

# Sort by the absolute score gap — biggest disagreement first
genre_analysis["abs_gap"] = genre_analysis["mean_gap"].abs()
genre_analysis = genre_analysis.sort_values("abs_gap", ascending=False)

print("Genre-level rating comparison:")
print(genre_analysis.drop("abs_gap", axis=1).to_string(index=False))

# Film-level: which single movie has the largest gap?
print("\nMost divisive individual films (by score gap):")
divisive = df[["title","genre","critic_score","audience_score","score_gap"]]
divisive = divisive.sort_values("score_gap", key=abs, ascending=False)
print(divisive.head(5).to_string(index=False))

# Compute Pearson correlation across all numeric columns
# This gives us critic vs audience, plus runtime and year as potential confounders
corr_matrix = df[["critic_score","audience_score",
                   "runtime_min","release_year"]].corr().round(3)

print("Full correlation matrix:")
print(corr_matrix.to_string())

# Extract just the critic-audience correlation for the business answer
ca_corr = corr_matrix.loc["critic_score","audience_score"]
print(f"\nCritic vs Audience correlation: {ca_corr:.3f}")

# Interpret strength
if abs(ca_corr) >= 0.7:
    strength = "strong — the two scores move together reliably"
elif abs(ca_corr) >= 0.4:
    strength = "moderate — some alignment but significant divergence"
else:
    strength = "weak — the two scores are largely independent"
print(f"Interpretation: {strength}")

# Check runtime and year correlations with both score types
print("\nCorrelations with runtime_min:")
print(f"  critic_score:   {corr_matrix.loc['critic_score','runtime_min']:.3f}")
print(f"  audience_score: {corr_matrix.loc['audience_score','runtime_min']:.3f}")

print("\nCorrelations with release_year:")
print(f"  critic_score:   {corr_matrix.loc['critic_score','release_year']:.3f}")
print(f"  audience_score: {corr_matrix.loc['audience_score','release_year']:.3f}")

# Top 3 films by critic score
print("Top 3 by critic score:")
top_critic = df.nlargest(3, "critic_score")[["title","genre","critic_score","audience_score","score_gap"]]
print(top_critic.to_string(index=False))

# Top 3 films by audience score
print("\nTop 3 by audience score:")
top_audience = df.nlargest(3, "audience_score")[["title","genre","critic_score","audience_score","score_gap"]]
print(top_audience.to_string(index=False))

# Films where the two systems diverge most — gap above 15 or below -10
print("\nHigh-divergence films (|gap| > 12):")
outliers = df[df["score_gap"].abs() > 12][
    ["title","genre","critic_score","audience_score","score_gap"]
].sort_values("score_gap", ascending=False)
print(outliers.to_string(index=False))

# Overall agreement rate: % of films within 10 points of each other
within_10 = (df["score_gap"].abs() <= 10).mean()
print(f"\nFilms within 10 points of each other: {within_10:.0%}")