Embedded Files
Highlights
Highlights
A/B testing is a user experience research methodology. A/B tests consist of a randomized experiment with two variants, A and B., which are identical except for one variation that might affect a user's behavior. It includes application of statistical hypothesis testing or "two-sample hypothesis testing" as used in the field of statistic
Approaches
We use hypothesis testing to test these two hypotheses:
Null Hypothesis: There is no difference in brand awareness between the exposed and control groups in the current case.
Alternative Hypothesis: There is a difference in brand awareness between the exposed and control groups in the current case.
Carried out 3 types of classification analysis to predict whether a user responds yes to brand awareness, namely: Logistic Regression Decision Trees XGboost, then compared the different classification models to assess the best performing one(s).
skills
Statistical Modelling
Using core data science python libraries pandas, matplotlib, seaborn, scikit-learn
ML algorithms Linear regression, Decision Trees, XGBoost
Model management (building ML catalog contains model feature labels and training model version)
MLOps with DVC, CML, and MLFlow
LIdar_3Dem
LIdar_3Dem
Python package that is used to fetch and process data from Amazon s3 bucket. LIDAR_3DEM package is built to fetch raster data from USGS Public dataset using a data pipeline. The package will process the raster data that is found in image format to geopandas dataframe that contains year aggregated coordinate points for a give bound. The coordinate points and their elevation will be plotted into a 3D image.
Approaches
Budil data pipeline to interact with API
Fetch satellite image data process, process, and transform it.
Provide python package
Skills
Explore satellite imagery as well as geographical data files
API interaction with remote storage
Code packaging and modularity
Building data pipelines and orchestrations workflows
Rossman Sales prediciton
Rossman Sales prediciton
Time series analysis on Rosemann pharmaceutical sales data across multiple stores. The project uses different machine learning algorithms to predict sales.
Approaches
Time series analysis explores the behavior of customers in the various stores, explores the contribution of number of customers and promotion on sales, seasonality of the purchase behavior.
Time-series forecasting Prediction of sales in various stores up to 6 weeks ahead of time using Xgboost, Logistic regression, and random forest regressor. Perform Random Forest Regressor to extract important features for sales prediction select top most important.
Using Mlfow and Dvc to track model and version data.
Skills
Python programming, working, Streamlit, Exploratory data analysis, Modeling, DVC, MLFlow, and CML.
Amharic Speech to text
Amharic Speech to text
The project aims to deliver an end-to-end Amharic speech recognition model using Tensorflow, keras, mlflow, and docker. the project will use Amharic speech data and transcribe it to text. For feature extraction, we have used MFCC. The project uses a combination of RNNand CNN algorithms for training and inference.
Approaches
Analyze the exiting speech data and transcription.
Map the audio file with transcription using Metadata.
Explore the audio and transcription.
Version data into a different form by augmentation techniques.
Extract features using MFCC.
Speech to text model was built using the combination of RNN and CNN tracking model using mlflow.
Skills
Audio and text processing, Deep learning methods (TensorfFlow, Keras, Pytorch),DVC, MLFLow, and CML.
Page updated
Google Sites
Report abuse