filmov
tv
How to create minimum viable product for machine learning projects - Weather prediction
Показать описание
Creating MVP for ML projects is an interesting topic because of quick feedback it provides for engaged partners (managers, clients, etc.) and can catch problems early in the process of product development. These feedbacks can also be used in improving next versions of product. In this video, I will show how a sample project can be analysed and then converted to a MVPs. I have kept it simple to focus on the main idea of starting with Jupyter notebook (data science side) and convert it to python (engineering side). There are many steps that can be added to this project and I will list some of the in the Jupyter notebook for you reference but if you think there is an important missing step in the process, please comment in the section below.
Codes:
🕒 VIDEO SECTIONS 🕒
0:00 - Intro
0:19 - Problem Definition
2:14 - Importing Data
4:46 - Changing data types - to_datetime
5:48 - Changing data types - LabelEncoder
8:28 - Reindexing - set_index
9:47 - Converting time series to conventional ML problem by shifting dataframe
18:55 - Model training
23:28 - Model evaluation
28:00 - Creating python files for MVP
Codes:
🕒 VIDEO SECTIONS 🕒
0:00 - Intro
0:19 - Problem Definition
2:14 - Importing Data
4:46 - Changing data types - to_datetime
5:48 - Changing data types - LabelEncoder
8:28 - Reindexing - set_index
9:47 - Converting time series to conventional ML problem by shifting dataframe
18:55 - Model training
23:28 - Model evaluation
28:00 - Creating python files for MVP
Комментарии