machine-learning-cheat-sheet.md

Machine Learning

:TODO: python framework mataflow from metaflow import FlowSpec, step :TODO: python framework sklearn from sklearn.pipeline import Pipeline

graph LR;
    A[<b>Data In</b>
    * filtered
    * cleaned
    * labeled
    ] --> 
    B(<b>ML Algorithm</b>
    * frameworks
    * algorithms
    🔄️
    );

    B --> 
    C(<b>Data out</b>
    ✅️ example 🆗️
    )    

Loading

---

graph 

m[<b>model</b>]
t[training]
i[inference]
t --associate--> m
i --associate--> m

r[regression
  model]
c[classification
  model]
c --extend--> m
r --extend--> m

l[label]
l --assign 
    to -->m

id[input data]
f[feature]
f --o id

idl[ <b>input data</b>
     labeled
     for training]
idnl[<b>input data</b>
    not labeled
    for prediction]
idl --extend--> id

idnl --extend--> id 
l --o idl

id ~~~ i

Loading

Necessary knowledges

graph LR;
    d[design] --> md[model <br>development] --> o[operations]
    md --> d 
    o --> md

Loading

design

• Requirement engineering
• ML UseCases prioritization
• Data Availability Check

model development

• Data Engineering
• ML Model Engineering
• Model Testing & Validation

operations

• ML Model Deployment
• CI/CD pipelines
• Monitoring & triggering

Frameworks

• 'double' ML pip install econml
• Google Jupyter Notebook