30a Coding a Naive Bayes classifier by hand
Welcome back to our Materials Informatics series! In this video, we continue exploring Bayesian and probabilistic machine learning by diving deeper into Naive Bayes classification.
In this video, we cover:
A brief recap of Naive Bayes, highlighting its simplicity and the assumption of feature independence.
A practical demonstration of Naive Bayes using materials project data to classify whether a material is a metal or an insulator.
How to fetch and prepare data from the materials project, including setting up API keys and defining criteria for data retrieval.
Calculation of mean and standard deviation for properties like density, volume, and formation energy for metals and insulators.
Visualization of data as probability distribution functions (PDFs) to understand the distribution of different material properties.
Application of Naive Bayes to classify a mystery material based on its density, volume, and formation energy.
Combining individual probability assessments to determine the overall likelihood of the material being a metal or an insulator.
A discussion on the limitations of Naive Bayes and the importance of handling unseen data points.
Stay tuned for our next video, where we explore Gaussian processes, a more sophisticated approach to Bayesian statistics in materials science.
Chapters:
00:00 Introduction and Recap
00:21 Naive Bayes and Feature Independence
00:42 Data Preparation and API Setup
01:36 Data Retrieval and Analysis
03:13 Visualizing Probability Distribution Functions
05:12 Applying Naive Bayes to Classify Materials
08:00 Combining Probabilities and Final Classification
09:44 Limitations and Next Steps
#MaterialsInformatics #NaiveBayes #ProbabilisticMachineLearning #BayesianStatistics #MaterialScience #MachineLearning #DataScience #GaussianProcesses