Upload ChatAttachmentAnalysis.py

ChatAttachmentAnalysis.py

@@ -15,7 +15,8 @@ df['embedding'] = df['embedding'].apply(lambda x: [float(num) for num in x.strip
 
 # Split the data into features (X) and labels (y)
 X = list(df.embedding.values)
-y = ['avoide', 'avoida', 'avoidb', 'avoidc', 'avoidd', 'anxietye', 'anxietya', 'anxietyb', 'anxietyc', 'anxietyd']
+y = df[['avoide', 'avoida', 'avoidb', 'avoidc', 'avoidd', 'anxietye', 'anxietya', 'anxietyb', 'anxietyc', 'anxietyd']].values
+
 
 # Split data into training and testing sets
 X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
@@ -34,3 +35,17 @@ mae = mean_absolute_error(y_test, preds)
 print(f"Chat transcript embeddings performance: mse={mse:.2f}, mae={mae:.2f}")
 
 
+# Mean Squared Error (MSE) is a measure of how close a fitted line is to data points.
+# In the context of this task, a lower MSE means that our model's predicted attachment scores are closer to the true scores.
+# An MSE of 1.32 suggests that the average squared difference between the predicted and actual scores is 1.32. 
+# Since our scores are normalized between 0 and 1, this error could be considered relatively high, 
+# meaning the model's predictions are somewhat off from the true values.
+
+# Mean Absolute Error (MAE) is another measure of error in our predictions. 
+# It's the average absolute difference between the predicted and actual scores.
+# An MAE of 0.96 suggests that, on average, our predicted attachment scores are off by 0.96 from the true scores.
+# Considering that our scores are normalized between 0 and 1, this error is also quite high, indicating that 
+# the model's predictions are not very accurate. 
+
+# Both MSE and MAE are loss functions that we want to minimize. Lower values for both indicate better model performance. 
+# In general, the lower these values, the better the model's predictions are.