Flashcards for TV Recommendation System Project

Q1: What libraries are imported at the start of the project?

A1: pandas, numpy, matplotlib.pyplot, TfidfVectorizer, cosine_similarity, linear_kernel.

---

Q2: Why is pd.set_option("display.max_colwidth", 200) used?

A2: To show full text (no truncation) when displaying columns like overviews or taglines.

---

Q3: How is the dataset loaded efficiently?

A3: By selecting only needed columns (name, overview, tagline, up to 8 genres, and numeric stats).

---

Q4: What does fillna("") do during preprocessing?

A4: Replaces missing values with empty strings for safe concatenation.

---

Q5: How are genres combined into one field?

A5: By joining all genres[i].name columns into a single string called genres_combined.

---

Q6: What is the purpose of the content column?

A6: It concatenates overview + tagline + genres into one text field for text analysis.

---

Q7: How is genre distribution analyzed?

A7: By tokenizing genres_combined, counting occurrences, and plotting the top 20 genres.

---

Q8: Which numeric features are used for correlation analysis?

A8: vote_average, vote_count, popularity, number_of_episodes, number_of_seasons.

---

Q9: How is the correlation heatmap created?

A9: By converting numeric columns to numbers, computing .corr(), and visualizing with a heatmap.

---

Q10: What does TfidfVectorizer(stop_words="english") do?

A10: Converts text into numerical vectors based on word importance, removing common stop words.

---

Q11: What is stored in tfidf_matrix?

A11: The TF-IDF representation of all shows’ content text.

---

Q12: Why is an indices mapping created?

A12: To quickly map show names to their row index in the dataframe.

---

Q13: How does the basic recommend function work?

A13:

1. Look up show index.

2. Compute cosine similarity with all shows.

3. Sort and select top N results (excluding itself).

---

Q14: What improvement is added in the second recommend function?

A14: A fuzzy search fallback that suggests close matches if the title is not found.

---

Q15: How are recommendations visualized?

A15: With a horizontal bar chart showing the cosine similarity of the top similar shows.