Changes

app.py

@@ -6,7 +6,7 @@ from sklearn.cluster import KMeans
 import matplotlib.pyplot as plt
 import seaborn as sns
 import plotly.express as px
-from mlxtend.frequent_patterns import apriori, association_rules
+from gensim.models import Word2Vec
 
 # Set the page configuration
 st.set_page_config(page_title="Customer Segmentation and Product Recommendation", layout="wide")
@@ -131,43 +131,47 @@ fig_cluster = px.scatter_3d(
 st.plotly_chart(fig_cluster)
 
 # Product Recommendation
-st.header("🛍️ Product Recommendation")
-st.sidebar.subheader("Recommendation Parameters")
-cluster_to_recommend = st.sidebar.selectbox("Select Cluster", rfm["Cluster"].unique())
-
-# Filter data by cluster
-customers_in_cluster = rfm[rfm["Cluster"] == cluster_to_recommend]["CustomerID"]
-df_cluster = df[df["CustomerID"].isin(customers_in_cluster)]
-
-# Association Rule Mining for Recommendations
-basket = (
-    df_cluster.groupby(["InvoiceNo", "Description"])["Quantity"]
-    .sum()
-    .unstack()
-    .fillna(0)
-    .applymap(lambda x: 1 if x > 0 else 0)
-)
-
-# Generate frequent itemsets
-frequent_itemsets = apriori(basket, min_support=0.05, use_colnames=True)
-
-# Generate association rules
-if not frequent_itemsets.empty:
-    rules = association_rules(frequent_itemsets, metric="lift", min_threshold=1)
-
-    # Display top recommendations
-    st.write(f"### Recommendations for Cluster {cluster_to_recommend}")
-    top_recommendations = rules.sort_values(by="confidence", ascending=False).head(10)
-    st.write(top_recommendations[["antecedents", "consequents", "support", "confidence", "lift"]])
-else:
-    st.write("No significant patterns found for this cluster.")
-
-st.write(f"### Recommendations for Cluster {cluster_to_recommend}")
-if not rules.empty:
-    top_recommendations = rules.sort_values(by="confidence", ascending=False).head(10)
-    st.write(top_recommendations[["antecedents", "consequents", "support", "confidence", "lift"]])
-else:
-    st.write("No significant patterns found for this cluster.")
+st.header("🎯 Product Recommendations")
+
+# Train Word2Vec Model
+st.subheader("🔍 Train Word2Vec Model")
+with st.spinner("Training Word2Vec model..."):
+    invoices = df.groupby("InvoiceNo")["Description"].apply(list)  # Group products by invoices
+    model = Word2Vec(sentences=invoices, vector_size=50, window=5, min_count=1, workers=4, sg=1)
+st.success("Word2Vec model trained successfully!")
+
+# Display similar products
+st.subheader("🔗 Find Similar Products")
+selected_product = st.selectbox("Select a product to find recommendations:", df["Description"].unique())
+
+if st.button("Recommend Products"):
+    try:
+        similar_products = model.wv.most_similar(selected_product, topn=5)  # Top 5 recommendations
+        st.write("### Recommended Products")
+        for product, similarity in similar_products:
+            st.write(f"- **{product}** (Similarity: {similarity:.2f})")
+    except KeyError:
+        st.warning("The selected product is not in the vocabulary. Please choose another.")
+
+# Recommendations for Cluster-Based Segmentation
+st.subheader("🔗 Recommendations by Cluster")
+cluster_to_recommend = st.selectbox("Select a cluster:", rfm["Cluster"].unique())
+
+if st.button("Recommend for Cluster"):
+    cluster_customers = rfm[rfm["Cluster"] == cluster_to_recommend]["CustomerID"]
+    cluster_df = df[df["CustomerID"].isin(cluster_customers)]
+    cluster_invoices = cluster_df.groupby("InvoiceNo")["Description"].apply(list)
+
+    with st.spinner("Training cluster-specific Word2Vec model..."):
+        cluster_model = Word2Vec(sentences=cluster_invoices, vector_size=50, window=5, min_count=1, workers=4, sg=1)
+
+    try:
+        cluster_similar_products = cluster_model.wv.most_similar(selected_product, topn=5)
+        st.write(f"### Recommended Products for Cluster {cluster_to_recommend}")
+        for product, similarity in cluster_similar_products:
+            st.write(f"- **{product}** (Similarity: {similarity:.2f})")
+    except KeyError:
+        st.warning("The selected product is not in the vocabulary for this cluster.")
 
 # Export Data
 st.header("📤 Export Processed Data")

requirements.txt

@@ -6,4 +6,4 @@ streamlit
 scikit-learn
 plotly
 tqdm
-mlxtend
+gensim