Python for AI #3: How to Train a Machine Learning Model with Python

preview_player
Показать описание
Python for AI Development Course #3: Today, we will learn how to train your first ML model with the data we prepared in the previous lesson. We will again use Jupyter Notebooks with Python and for the Machine Learning model training, we will use Scikit-learn.

Get your Free Token for AssemblyAI👇

00:00 Introduction
00:40 What can you do with Scikit-learn
03:25 Splitting the data into train and test
05:03 Imbalanced learn
07:28 Choosing the right ML model
08:47 Training the model
11:15 Evaluating the model
16:19 Model tuning
19:48 Cross-validation
22:00 Wrap-up

▬▬▬▬▬▬▬▬▬▬▬▬ CONNECT ▬▬▬▬▬▬▬▬▬▬▬▬

▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬

#MachineLearning #DeepLearning
Рекомендации по теме
Комментарии
Автор

Ever since I came across your videos, I can't stop watching. I've seen a lot on the internet, but you are the only ones who can present information so clearly. Thank U a lot!

olgav.
Автор

Great course. Just a thought, would be nice to add a #4 where you pick one and go through it together with us and slowly. Again, very well done!

kimaegaii
Автор

00:04 Training a machine learning model using scikit-learn
02:15 Scikit-learn provides useful metrics and pre-prepared datasets for machine learning.
04:29 Splitting Data into Train, Test, and Validation Sets
06:49 The main steps in the process are data set sampling, model training, and model selection
08:59 Scikit-learn documentation provides thorough explanations and examples of importing and using different algorithms and models.
11:10 There are multiple ways to evaluate machine learning models
13:25 Creating a confusion matrix and classification report using Seaborn and scikit-learn
15:42 Accuracy alone may not provide a complete picture of model performance.
17:46 GridSearchCV helps create and compare multiple models with different hyperparameters.
19:41 Grid search includes cross-validation for training models with different parameter combinations.
21:44 Get started with scikit-learn: data preparation, model training, and evaluation.
Crafted by Merlin AI.

sarthakmalik
Автор

Not a big deal since you didn't change any default parameters but you implemented a classifier while showing the regression documentation
Great high level overview of the ML training process.

benstallone
Автор

We got the setup, prepared the data, we did the training and the validation. But in the end we never got to use the model. I thought this would be the most important part.

metanulski
Автор

Hi @Misra,

I'm using the flights dataset, full as submited in Kaggle, I'm running the clf = followed by the cross_val_score and it stay for so long time processing and don't finish at least with 1:30 hour. I'm using a MacBook Pro M1 with 64GB, Is there anything that I'm missing?

amx
Автор

Where can I get the cheat sheet for choosing model

statusmart
Автор

Great video! And she is the most beautiful programmer I have ever seen ❤

DanielADamico
Автор

Not a step by step guide this time, I was able to follow along though the code errors just after In [44] as X_resampled is not defined. Interestingly my seaborn heatmap was mostly dark.

smudgepost
Автор

Thanks for the explanation mam it helped, i m having a doubt with my project....see as i working on a AI LLM module so I have completed the training the dataset and all that stuff....now i stucked at testing the module I m not getting that how to test your
Can you please guide over that....like maybe a proper video describing how to test you ML Models r AI LLM will help a lot mam!!....I need

adityatiwari
Автор

Well explained however I feel it was a little bit rushed.
I'm wondering why the model (shown in the classification_report) is still so unbalanced even after RandomOversamlper did rebalance the classes.
Anyway PCA and dimensionality reduction topics are welcomed even to complement this tutorial, given you've skip to encode the other non numerical data that could have provided good insight for the model and provide better AP

EliSpizzichino
Автор

why not simply show examples step by step, rather than explaining stuff we dont even do?

ttaylor