AI 2025 Datascience Foundational 44 projects Zero To Pro

All the files necessary for this section are in this zip file.  You must download the file, unzip it, then open the folder from within Jupyter.  

–Scikit AKA sklearn

–Load dataset

–Quantiles

–Describe() method

–Range

–Aggregate()

–Confidence Intervals

–Confidence Interval for Population Proportion

–Testing the Proportion and chi squared

–Two- Sample Test for Common Mean

– One way ANOVA

Learn about Bayes Theorem and how to use jupyter.

–Compare two proportions

–Posterior Analysis

–Hypothesis Testing

–Conjugate Priors

–Conjugate Priors

–DNIG

–Credible Interval for the Mean

–Pearson correlation

–Using Pandas and SciPy

–Sklearn with Dataset

–Matplots heatmap

We will cover:

–Supervised vs Unsupervised Learning

–Matplots

–Supervised Learning

–Loss Function

–Generalize

–Testing Algorithm predictions

–Overfitting

–Underfitting

–Cross Validation

–k Folds

–Using Dataframe Methods (.mean())

–Precision

–Recall

–F1 Score

– –Bayes Factor

All the notes for this section are in this folder.  Download here then you can use these in Jupyter to go through the lessons.  

– k – Nearest neighbors algo

–Creating Classifiers

–Selecting k

–Using k to predict 

–Scikit – learn

–Fitting a Decision Tree

–Control overfitting by restricting  tree depth

–Ensemble methods

–Growing  a Random Forest

–Optimizing Hyperparameters

–Linear Models with OLS

–Regression

–Fitting

–Cross validation

–Statsmodel for fitting linear models

–Use AIC

–Comparing models using mean squared Error

–Least absolute shrinkage and selection operator

–Fitting using LASSO

–Using smaller alpha for better fits

–Compare to OLS

– spline interpolation

–Univariate interpolation 1D

–Multivariate interpolation 2D

–Training a perceptron type linear classifier

–Online learning

–Handwriting analysis

–Multilayer perceptron

–Fitting an MLP 

–Basic RNN with 100 neurons

–Alpha is 10

–Distance types

–Euclidiean

–Angular

–Hamming

–Jaccard

–Manhattan

–Clusters in datasets

–Finding elbow

–Silhouette Analysis 

–K means clustering

– image compression

–Hierarchal Clustering

–Similarity Scores

–Agglomerative Clustering

–Clustering Words

–Silhouette Plots

–Spectral clustering

–Clustering words

–Principal Component Analysis

–Dimensionality reduction when uncorrelated 

–Singular value decomposition

–Dense vs sparse data

–Compact SVD

–Find matrices

–Multi-dimensional scaling

–Reduce dimensionality then restore

–Sometimes no clustering restored

All the notes for this section, Machine Learning AI, are in this folder.  

Supervised Learning

Unsupervised Learning

Reinforcement Learning

–Artificial neurons, neural gates

–Perceptron

–Artificial neurons, neural gates

–Perceptron

–Gradient descent and feature scaling

–Choosing algos

–Perceptron models

–Vector machines

–Kernel SVN with hyperplanes

–kNN

–Decision tree

–Choosing algos

–Perceptron models

–Vector machines

–Kernel SVN with hyperplanes

–kNN

–Decision tree

–Dealing with missing data

–Scikit learn API

–Categorical data

–One hot encoding on nominal

–Geometric L1/L2 regularization

–Random forests

–Dealing with missing data

–Scikit learn API

–Categorical data

–One hot encoding on nominal

–Geometric L1/L2 regularization

–Random forests

–Dealing with missing data

–Scikit learn API

–Categorical data

–One hot encoding on nominal

–Geometric L1/L2 regularization

–Random forests

–PCA

–Variance total and explained

–LDA

–Kernel PCA

–PCA

–Variance total and explained

–LDA

–Kernel PCA

–PCA

–Variance total and explained

–LDA

–Kernel PCA

–Combining estimators and transformers

–Learning curves bias vs variance

–Plotting underfitting overfitting

–Algo selection 

–Combining estimators and transformers

–Learning curves bias vs variance

–Plotting underfitting overfitting

–Algo selection 

–Combining estimators and transformers

–Learning curves bias vs variance

–Plotting underfitting overfitting

–Algo selection 

–Unanimity vs majority vs plurality

–Majority voting to improve predictions

–Bagging

–Leveraging weak learners via adaptive boosting 

–Unanimity vs majority vs plurality

–Majority voting to improve predictions

–Bagging

–Leveraging weak learners via adaptive boosting 

–Preprocessing and cleaning data

–Word relevancy via frequency inverse document frequency

–Documents to tokens

–Training regression model for doc classification

–Latent Dirichlet allocation

–Preprocessing and cleaning data

–Word relevancy via frequency inverse document frequency

–Documents to tokens

–Training regression model for doc classification

–Latent Dirichlet allocation

–Unanimity vs majority vs plurality

–Majority voting to improve predictions

–Bagging

–Leveraging weak learners via adaptive boosting 

–Linear regression simple

–Multiple linear regression

–Visulize all dataset

–Linear regression simple

–Multiple linear regression

–Visulize all dataset

–K means using scikit

–Hierarchical tree clusters

–DBSCAN      

–K means using scikit

–Hierarchical tree clusters

–DBSCAN      

–Multilayer neural network

–Handwriting analysis

–Training ANN, logistic cost function

–Neural backpropagation

–Multilayer neural network

–Handwriting analysis

–Training ANN, logistic cost function

–Neural backpropagation

–Learning models with TensorFlow

–Build MNN w/ Tensor APIs 

–Choosing activation functions MNNs

–Learning models with TensorFlow

–Build MNN w/ Tensor APIs 

–Choosing activation functions MNNs

–Learning models with TensorFlow

–Build MNN w/ Tensor APIs 

–Choosing activation functions MNNs

–Ranks

 –Tensors to multidimensional arrays

–Computation graph

–Tensor board 

–Ranks

 –Tensors to multidimensional arrays

–Computation graph

–Tensor board 

–CNN padding, convolutional output

–Training w/dropout

–Tensorboard for CNN

–Weights

–Biases

–Reduce overfitting  

–CNN padding, convolutional output

–Training w/dropout

–Tensorboard for CNN

–Weights

–Biases

–Reduce overfitting  

–CNN padding, convolutional output

–Training w/dropout

–Tensorboard for CNN

–Weights

–Biases

–Reduce overfitting  

–Single to multi RNN

–Training and long range interactions

–Build language translator

–Character level modeling

–Learning over 200 epochs

–Single to multi RNN

–Training and long range interactions

–Build language translator

–Character level modeling

–Learning over 200 epochs

–Single to multi RNN

–Training and long range interactions

–Build language translator

–Character level modeling

–Learning over 200 epochs

–Single to multi RNN

–Training and long range interactions

–Build language translator

–Character level modeling

–Learning over 200 epochs