Coursera - Tools for Data Science and Machine Learning Model Deployment

This repository contains the materials, source codes and links which were explored and experimented for "Data Science and Machine Learning" Courses as part of my Ph.D. Coursework (Oct, 2021 - Jan, 2023).

Course - 1 : Tools for Data Science

IBM Watson Studio - Jupyter Notebook Assignment
IBM Watson Studio - Jupyter Notebook Code

Languages and Library & Tools of Data Science

• Python
  • Pandas : Data Structure & Tools
  • numpy : Arrays & Matrices
  • Scipy
  • matplotlib
  • PyTorch : For Experimentation and Test Research Ideas
  • TensorFlow : Production and Deployment
  • Keras : Deep Learning Neural Networks
  • Scikit-learn : Tools for Statistical Modeling including Regression, Classification and Clustering
• R
• SQL
• Java
  • Weka : Data-mining
  • Java-ML : Machine Learning
  • Apache MLlib : Scalable ML
  • Deeplearning4j
  • Hadoop
• Scala
  • Spark
• C++
  • TensorFlow
  • MongoDB
  • Caffe
• JavaScript
  • TensorFlow.js
  • R.js
  • brain.js
  • machinelearn.js
• Julia
  • JuliaDB

Module based Tools

• Data Management
  • Open Source
    • Relational DB : MySQL and PostgreSQL
    • File-based : Hadoop
    • NoSQL DB : MongoDB, CouchDB, Apache Cassandra
    • Cloud File : Ceph
  • Commercial
    • Oracle Database
    • Microsoft SQLServer
    • IBM DB2
  • Cloud
    • AWS Amazon Dynamo DB
    • Cloudant
    • IBM DB2
• Data Integration and Transformation
  • Open Source
    • Apache AirFlow
    • Apache NiFi
    • KubeFlow
    • Spark SQL
    • Apache Kafka
    • Node-RED
  • Commercial
    • Informatica
    • IBM Infosphere Datastage
    • Watson Studio Desktop
    • Talend
  • Cloud
    • Informatica
    • IBM Data Refinery
• Data Visualization
  • Open Source
    • Hue
    • Kibana
    • Apache Superset
  • Commercial
    • Tableau
    • Microsoft Power BI
    • IBM Cognos
  • Cloud
    • Datameer
    • IBM Cognos Analytics
• Model Building
  • Commercial
    • IBM SPSS Modeler
    • SAS Enterprise Miner
• Model Deployment
  • Open Source
    • Prediction10
    • Seldon
    • MLeap
    • TensorFlow Service
  • Cloud
    • IBM Watson ML
• Model Monitoring and Assessment
  • Open Source
    • ModelDB
    • Prometheus
    • IBM AI Fairness 360 OpenSource
    • IBM Adversarial Robutness 360 Toolbox
    • IBM AI Explainability 360 Toolkit
  • No Major Commercial Products
  • Cloud
    • AWS Amazon SageMaker Model Monitor
• Code Asset Management
  • Open Source
    • Git : Gitlab, GitHub
    • Bitbucket
• Data Asset Management
  • Open Source
    • Apache Atlas
    • ODPI EGeria
    • Kylo
  • Commercial
    • Informatica
• Development Environment
  • Open Source
    • Jupyter
    • Jupyter Lab
    • Zeppelin Notebook
    • R Studio
    • Spyder
  • Commercial
    • Watson Studio Desktop
• Cluser Execution Environment
  • Open Source
    • Apache Spark
    • Apache Flink
    • Ray
• Fully Integrated Visual Tools
  • Open Source
    • KNIME
    • Orange
  • Commercial
    • IBM Watson Studio
    • IBM Watson OpenScale
    • H20 Driverless AI

Dataset Sources

• Open Knowledge Foundation
• Kaggle
• Google Data Set Search
• IBM Data Asset eXchange (DAX)

Course - 2 : Deploying Machine Learning Models in Production

• Model Serving

  • Input Feature Lookup

    • Google Cloud Memorystore
    • Google Cloud Firestore
    • Google Cloud Bigtable
    • Amazon DynamoDB

  • Model Deployment Servers

    • TensorFlow Lite
    • Clipper
    • TensorFlow Serving
    • KubeFlow Serving
    • NVIDIA Triton Inference Server

• Data Versioning

  • Neptune
  • Pachyderm
  • Delta Lake
  • Git LFS
  • Dolt
  • LakeFS
  • DVC
  • ML-Metadata
  • Wandb

• Logging Metrics

  • TensorBoard

• Notebooks

  • nbconvert
  • nbdime
  • jupytext
  • neptune-notebooks
  • git

Lab Exercises - Google Cloud

• Autoscaling TensorFlow model deployments with TF Serving and Kubernetes
• TFX on Google Cloud Vertex Pipelines
• Implementing Canary Releases of TensorFlow Model Deployments with Kubernetes and Anthos Service Mesh
• Data Loss Prevention: Qwik Start - JSON