BookRecommendation / src /streamlit_app.py
EnYa32's picture
Update src/streamlit_app.py
e521045 verified
Raw History Blame Contribute Delete
1.75 kB
import streamlit as st
import pandas as pd
import joblib
from scipy import spatial
st.set_page_config(page_title="Book Recommender", page_icon="๐Ÿ“š", layout="wide")
@st.cache_resource
def load_all():
df = pd.read_csv("src/bookss_data.csv")
df["average_rating"] = df["average_rating"].astype(float)
df["content"] = df["title"] + " " + df["authors"]
tf = joblib.load("src/tfidf_vectorizer.pkl")
X = tf.transform(df["content"])
return df, tf, X
df, tf, X = load_all()
def compute_distance(i, j, rating_weight=0.3):
v1 = X[i].toarray().ravel()
v2 = X[j].toarray().ravel()
content_distance = spatial.distance.cosine(v1, v2)
rating_distance = abs(df.loc[i, "average_rating"] - df.loc[j, "average_rating"]) * rating_weight
return content_distance + rating_distance
def recommend(title, top_n=10, rating_weight=0.3):
idx = df.index[df["title"] == title][0]
distances = []
for j in range(len(df)):
if j == idx:
continue
d = compute_distance(idx, j, rating_weight=rating_weight)
distances.append((df.loc[j,"bookID"], df.loc[j,"title"], df.loc[j,"average_rating"], d))
distances.sort(key=lambda x: x[3])
return distances[:top_n]
st.title("๐Ÿ“š Book Recommendation System")
selected_title = st.selectbox("Choose a book", sorted(df["title"].unique()))
top_n = st.slider("Top N", 3, 20, 10)
rating_weight = st.slider("Rating weight", 0.0, 1.0, 0.3, 0.05)
if st.button("Recommend"):
recs = recommend(selected_title, top_n=top_n, rating_weight=rating_weight)
rec_df = pd.DataFrame(recs, columns=["bookID","title","average_rating","distance"])
st.dataframe(rec_df, use_container_width=True)
st.caption("Smaller distance = more similar.")