Guided Project: Build a Sentiment Analyser in PyCharm
Train a machine learning model that tells positive reviews from negative ones, measure how good it really is, and find where it goes wrong.
Guided project, Beginner · Machine Learning · about 1.5 hours · 1 module · certificate on completion
Taught by Khalidà Sultàn, AI Learning Companion.
About this course
In this guided project you build a working sentiment analyser from scratch in PyCharm: a model that reads a short product review and says whether it's positive or negative. Three short lessons explain the ideas you need, how TF-IDF turns words into numbers, how logistic regression learns which words matter, and how to tell whether a classifier is any good, each shown step by step in PyCharm's Python Console, editor and debugger. Then you build it: you load 40 labelled reviews written for the project, train a scikit-learn pipeline, test it on reviews it hasn't seen, check it with cross-validation, look at the words it relies on, and try it on sentences of your own, including the ones that fool it. You finish with a model you understand, and a clear sense of what a small text classifier can and can't do.
What you'll learn
- Turn short texts into numbers with TF-IDF
- Train a logistic regression sentiment classifier in a scikit-learn pipeline
- Evaluate it with a test set, a confusion matrix and cross-validation
- Test it on new sentences and explain where and why it fails
Skills you'll gain
Sentiment analysis, Text classification, TF-IDF, Logistic regression, scikit-learn, Model evaluation, Cross-validation
Tools
PyCharm, Python, scikit-learn
Before you start
Basic Python: variables, lists, functions and running a script. PyCharm (the free version is enough) and Python 3 installed; you'll install scikit-learn in the first lesson.
Modules
Build, evaluate and test a sentiment analyser
Learn how text becomes numbers and how a classifier learns from them, then build a sentiment analyser, measure it honestly and probe its limits.
- From words to numbers: bag of words and TF-IDF (8 min)
- Training a classifier: logistic regression in a pipeline (10 min)
- Is it any good? Evaluating a classifier, and its limits (10 min)
- Hands-on lab: Build and test a sentiment analyser
PyCharm is a trademark of JetBrains s.r.o. This course is independent: JetBrains hasn't made or endorsed it.