Product length prediction on the basis of product title , description and product type ID
Welcome to our project, "Achieving Success in the Amazon ML Challenge: A Journey of Team NazarickAI"! In this repository, we share our approach, techniques, and the journey that led us to secure the 74th position in the Amazon ML Challenge among teams from across India. Our team, comprised of CH. Satya Savith, Akshat Jain, Ankit Dubey, and myself, embarked on this challenging adventure and achieved remarkable results.
The Amazon ML Challenge presented us with the task of determining the length of a product based on its title, tags, description, and product type ID. We were provided with a dataset that contained both textual and numeric features. To solve this problem, we aimed to leverage the information contained in the product details (textual data) and the product type ID (numeric data) to accurately predict the length of the product.
Our approach to the challenge consisted of the following key steps:
-
Thorough Data Analysis: We conducted a comprehensive analysis of the dataset, gaining insights into the problem's nature and understanding the relationships between features.
-
Text Vectorization: To handle the textual data, we employed the term frequency-inverse document frequency (TF-IDF) technique. TF-IDF allowed us to assign weights to words based on their importance in the context, enabling us to capture valuable information from the product title, tags, and description.
-
Incorporating Numeric Features: Recognizing the significance of the product type ID as a numeric feature, we included it in our predictive model. This fusion of text and numeric features helped create a robust and accurate solution.
-
Model Selection: We explored various machine learning algorithms, including KMeans clustering, K-nearest neighbors (KNN), XGBoost, linear regression, deep neural networks (DNN), gradient boosting (GB), stochastic gradient descent (SGD), long short-term memory (LSTM), decision tree, and AutoML.
To get started with our project, follow these steps:
- Clone the repository using the following command: git clone https://github.com/heathbrew/Amazon-ML-hackathon-Product-length-prediction.git