Upload 5 files

app.py

@@ -0,0 +1,118 @@
+import streamlit as st
+import pandas as pd
+import numpy as np
+import lightgbm as lgb
+from datetime import datetime
+
+# Load the trained model
+model = lgb.Booster(model_file='lightgbm_model.txt')
+
+# Function to convert date to timestamp
+def date_to_timestamp(date):
+    return datetime(date.year, date.month, date.day).timestamp()
+
+# Function to create the input dataframe
+def create_input_df(date, visitNumber, totals_visits, totals_hits, totals_pageviews, totals_newVisits,
+                    channelGrouping, device_browser, device_operatingSystem, device_deviceCategory, 
+                    geoNetwork_subContinent, trafficSource_source, trafficSource_medium):
+    data = {
+        'date': [date_to_timestamp(date)],
+        'visitNumber': [visitNumber],
+        'totals_visits': [totals_visits],
+        'totals_hits': [totals_hits],
+        'totals_pageviews': [totals_pageviews],
+        'totals_newVisits': [totals_newVisits],
+        f'channelGrouping_{channelGrouping}': [1],
+        f'device_browser_{device_browser}': [1],
+        f'device_operatingSystem_{device_operatingSystem}': [1],
+        f'device_deviceCategory_{device_deviceCategory}': [1],
+        f'geoNetwork_subContinent_{geoNetwork_subContinent}': [1],
+        f'trafficSource_source_{trafficSource_source}': [1],
+        f'trafficSource_medium_{trafficSource_medium}': [1],
+    }
+
+    # Create the dataframe
+    input_df = pd.DataFrame(data)
+
+    # Fill missing columns with 0
+    expected_columns = [
+        'date', 'visitNumber', 'totals_visits', 'totals_hits', 'totals_pageviews', 'totals_newVisits',
+        'channelGrouping_Direct', 'channelGrouping_Organic Search', 'channelGrouping_Paid Search', 
+        'channelGrouping_Referral', 'channelGrouping_Social',
+        'device_browser_Android Webview', 'device_browser_Chrome', 'device_browser_Firefox', 
+        'device_browser_Internet Explorer', 'device_browser_Safari',
+        'device_operatingSystem_Android', 'device_operatingSystem_Chrome OS', 'device_operatingSystem_Linux', 
+        'device_operatingSystem_Macintosh', 'device_operatingSystem_Windows', 'device_operatingSystem_iOS',
+        'device_deviceCategory_desktop', 'device_deviceCategory_mobile', 'device_deviceCategory_tablet',
+        'geoNetwork_subContinent_Eastern Asia', 'geoNetwork_subContinent_Eastern Europe', 
+        'geoNetwork_subContinent_Northern America', 'geoNetwork_subContinent_Northern Europe', 
+        'geoNetwork_subContinent_South America', 'geoNetwork_subContinent_Southeast Asia', 
+        'geoNetwork_subContinent_Southern Asia', 'geoNetwork_subContinent_Southern Europe', 
+        'geoNetwork_subContinent_Western Asia', 'geoNetwork_subContinent_Western Europe',
+        'trafficSource_source_(direct)', 'trafficSource_source_google', 'trafficSource_source_mall.googleplex.com', 
+        'trafficSource_source_youtube.com', 'trafficSource_medium_(none)', 'trafficSource_medium_affiliate', 
+        'trafficSource_medium_cpc', 'trafficSource_medium_organic', 'trafficSource_medium_referral'
+    ]
+
+    for col in expected_columns:
+        if col not in input_df.columns:
+            input_df[col] = 0
+
+    return input_df[expected_columns]
+
+# Streamlit app
+st.title('GStore Digital Sales Revenue Prediction 🪝')
+st.image('google_store.jpg')
+st.write('Google shared some digital marketing data for their store. This app calculates whether a customer profile will turn into a sale or not.')
+
+st.write('High Page visits and site views, using Chrome OS, using mall.googleplex.com, located in North America are all have increased revenue possibility')
+
+# Collect user inputs
+date = st.date_input('Date')
+visitNumber = st.number_input('Visit Number', min_value=1)
+totals_visits = st.number_input('Total Visits', min_value=1)
+totals_hits = st.number_input('Total Hits', min_value=1)
+totals_pageviews = st.number_input('Total Pageviews', min_value=1)
+totals_newVisits = st.number_input('Total New Visits', min_value=0)
+
+channelGrouping = st.selectbox('Channel Grouping', 
+                               ['Direct', 'Organic Search', 'Paid Search', 'Referral', 'Social'])
+device_browser = st.selectbox('Device Browser', 
+                              ['Android Webview', 'Chrome', 'Firefox', 'Internet Explorer', 'Safari'])
+device_operatingSystem = st.selectbox('Device Operating System', 
+                                      ['Android', 'Chrome OS', 'Linux', 'Macintosh', 'Windows', 'iOS'])
+device_deviceCategory = st.selectbox('Device Category', ['desktop', 'mobile', 'tablet'])
+geoNetwork_subContinent = st.selectbox('GeoNetwork Sub Continent', 
+                                       ['Eastern Asia', 'Eastern Europe', 'Northern America', 'Northern Europe', 
+                                        'South America', 'Southeast Asia', 'Southern Asia', 'Southern Europe', 
+                                        'Western Asia', 'Western Europe'])
+trafficSource_source = st.selectbox('Traffic Source', ['(direct)', 'google', 'mall.googleplex.com', 'youtube.com'])
+trafficSource_medium = st.selectbox('Traffic Source Medium', ['(none)', 'affiliate', 'cpc', 'organic', 'referral'])
+
+# Create input dataframe
+input_df = create_input_df(date, visitNumber, totals_visits, totals_hits, totals_pageviews, totals_newVisits,
+                           channelGrouping, device_browser, device_operatingSystem, device_deviceCategory, 
+                           geoNetwork_subContinent, trafficSource_source, trafficSource_medium)
+
+# Make prediction
+prediction = model.predict(input_df)
+
+# Display the prediction result
+if prediction[0] < 0:
+    st.header("The user is not likely to buy.")
+elif prediction[0] < 0.001:
+    st.header("The user is not likely to buy.")
+elif prediction[0] < 0.02:
+    predicted_value = prediction[0] * 5000
+    st.header(f"The user might buy. Predicted Revenue: ${int(predicted_value)}")
+else:
+    predicted_value = prediction[0] * 2500
+    st.header(f"The user is highly likely to buy. Predicted Revenue: ${int(predicted_value)}")
+
+st.write('These are some graphs about the 900K rows of data. 1.3% of site visitors turn into customers')
+st.image('Unknown-1.png')
+st.write('The high spenders are what keeps the wheel going')
+st.image('Unknown.png')
+st.write('The visits and sales are quite corraleted and seasonal.')
+st.image('Unknown-2.png')
+