You will get End-To-End Machine Learning Model for your Dataset
Rising Talent
Project details
This service will train, validate & test Machine Learning models on your dataset, for Regression or Classification tasks.
We'll first perform EDA on your dataset. Pre-cleaning & Post-cleaning EDA in Standard & above tiers. Else just Post-cleaning. Missing values, de-duplication, outlier handling etc. To provide analysis of data before modelling. Basically the outputs of the Standard tier of my Data Cleaning & EDA project catalog service.
Then models trained on original raw data. Data Cleaning, Pre-Processing & Feature Engineering etc included in final pipeline model object.
Algorithms used, by tier:
Basic:
Regression: Linear Regression
Classification: Logistic Regression
Standard:
Basic + Decision Tree, Random Forest
Premium:
Standard + Gradient Boost, XGBoost, LightGBM
Standard tier includes Hyper-parameter Tuning of model with best Validation score.
Premium tier has tuning for up to 3 models.
Trained pipeline models will be packaged as Pickle files in Basic tier, also Joblib in Standard and Premium.
Deliverables include reports and analytics plots shown in the gallery plus JSON reports. Model Usage Guide included.
Let's connect & build the accurate ML models you need!
We'll first perform EDA on your dataset. Pre-cleaning & Post-cleaning EDA in Standard & above tiers. Else just Post-cleaning. Missing values, de-duplication, outlier handling etc. To provide analysis of data before modelling. Basically the outputs of the Standard tier of my Data Cleaning & EDA project catalog service.
Then models trained on original raw data. Data Cleaning, Pre-Processing & Feature Engineering etc included in final pipeline model object.
Algorithms used, by tier:
Basic:
Regression: Linear Regression
Classification: Logistic Regression
Standard:
Basic + Decision Tree, Random Forest
Premium:
Standard + Gradient Boost, XGBoost, LightGBM
Standard tier includes Hyper-parameter Tuning of model with best Validation score.
Premium tier has tuning for up to 3 models.
Trained pipeline models will be packaged as Pickle files in Basic tier, also Joblib in Standard and Premium.
Deliverables include reports and analytics plots shown in the gallery plus JSON reports. Model Usage Guide included.
Let's connect & build the accurate ML models you need!
Machine Learning Tools
Microsoft Excel, NumPy, pandas, Python, Python Scikit-Learn, scikit-learn, SciPy, XGBoostWhat's included
| Service Tiers |
Starter
$200
|
Standard
$400
|
Advanced
$800
|
|---|---|---|---|
| Delivery Time | 2 days | 4 days | 6 days |
Number of Revisions | 1 | 1 | 1 |
Number of Model Variations | 1 | 3 | 6 |
Number of Scenarios | 1 | 1 | 1 |
Number of Graphs/Charts | 3 | 5 | 7 |
Model Validation/Testing | |||
Model Documentation | |||
Data Source Connectivity | - | - | - |
Source Code | - | - | - |
Optional add-ons
You can add these on the next page.
Fast Delivery
+$50 - $200
Additional Revision
+$200
AutoML
+$100Frequently asked questions
About Hammad
Data Scientist | Machine Learning & Data Services
Rawalpindi, Pakistan - 8:58 pm local time
• Data Cleaning services such as De-duplication, Missing value handling (Interpolation, forward-fill, back-fill, KNN & other Impute methods), Outlier handling (Z-Score Winsorized, IQR method).
• Exploratory Data Analysis including Correlation heat-maps, data distribution charts (hist, scatter, bar), Missing data & Box plots and summary of findings.
• Time-Series Analysis including Autocorrelation test plots (ACF, PACF), Stationarity tests (ADF, KPSS) and Trend Decomposition.
• Data Drift detection via KS_Statistic and p_values.
• Feature Engineering and Data Pre-Processing in preparation for Machine Learning. Creation of new features from existing numerical, categorical, and datetime features. Scaling and Encoding of numerical and categorical data. Class Balancing etc.
• Feature Selection via Mutual Info method, variance of feature with target, filtering for minimal covariance with other features and multi-collinearity etc. Stratified splitting and producing the final set of datasets in the form of Training-Validation-Testing splits.
• Building of Machine Learning models using the cleaned and processed data from above steps. Training multiple models using different algorithms and comparing the results. Algorithms used depend on whether the task is Regression (Linear Regression, Ridge, Lasso, Elastic Net, AutoGluon models etc) or Classification (Decision Trees, Random Forests, XGBoost, FLAML etc).
• Hyper-parameter Tuning via Random Search or Optuna Bayesian etc.
• Post modelling validation and testing using split data from previous steps. Comparisons of metrics scores between splits and algorithms.
• Post-Modelling Visualizations such as Feature Importances plots, Metrics Tables, Predicted VS Actuals plots, Residuals Analysis, Classification Reports, Precision Recall Curves, SHAPley plots etc.
And much more.
I've worked as a data scientist at multinational corporations such as AkzoNobel and British American Tobacco. I worked in a data analysis capacity in both of these roles, as well as building machine learning models using Python for various predictive tasks.
Specific examples of these tasks:
Using Neural Networks to build a model that predicts the quality of a type of paint based on the chemical composition. Then building a Genetic Algorithm based system that uses said neural network model to generate new paints as chemical compositions, that are predicted to be high quality.
Improving a machine learning model used to predict price elasticities for cigarette products, iterating upon original Linear Regression version by instead using XGB Regressor, Random Forests, ARIMA & SARIMAX etc. Then using the models to build a price scenario tool in Excel to make sales forecasts and predict revenue based on different price scenarios.
If you feel you might be in need of any of my services, please feel free to drop me a message so we can discuss. I've also uploaded services that I offer here on Upwork in my Project Catalog, each of which has a relevant item in my portfolio. Please check these out for an idea of what exactly I offer as final deliverables for each service.
Thanks for reading this, and I hope you'll reach out. Have a nice day!
Steps for completing your project
After purchasing the project, send requirements so Hammad can start the project.
Delivery time starts when Hammad receives requirements from you.
Hammad works on your project following the steps below.
Revisions may occur after the delivery date.
Gathering Inputs
I will gather the input dataset and preferences from the client.
Setup and Run Pipelines
I will setup my Python pipelines using the provided data and settings, and then will run said pipelines to completion.