Learn core machine learning algorithms, supervised and unsupervised learning, feature engineering and model evaluation with practical Python examples. Work through the ML workflow examples in order, then change one assumption and explain whether your interpretation still holds. This page targets the practical intent behind ML workflow, not repetition of the phrase itself.
Editorial note: Publication metadata for Machine Learning Tutorial is intentionally unset until this URL is actually released; the lesson already includes its own examples, CSV and notebook.
Frame the prediction problem before choosing a model
For loan default prediction, you might compare logistic regression with a random forest using a time-aware split, inspect precision and recall at multiple thresholds, and choose a threshold based on the cost of missed defaults versus unnecessary reviews.
- Machine learning learns patterns from examples, but the quality of labels, split strategy and evaluation often matters more than model complexity.
- A baseline model provides the reference needed to decide whether a sophisticated algorithm adds real value.
- Training performance alone is not evidence of generalization; evaluation data must remain separate from model fitting and tuning.
- Error analysis should identify where the model fails so the next iteration targets a real weakness instead of random hyperparameter changes.
Build a baseline and a leakage-safe evaluation
For ML workflow, predict one visible result before each snippet runs—a row count, shape, statistic, ranking or error pattern—then compare it with the printed checkpoint.
# ML workflow · example 1
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import classification_report
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=.2, stratify=y, random_state=42)
model = LogisticRegression(max_iter=1000).fit(X_train, y_train)
print(classification_report(y_test, model.predict(X_test)))ML workflow checkpoint 1: Run the cell and inspect the printed values before changing anything.
# ML workflow · example 2
import pandas as pd
from sklearn.model_selection import train_test_split
df = pd.read_csv('practice.csv')
target = df.select_dtypes('number').columns[-1]
X = df.drop(columns=[target]).select_dtypes('number')
y = df[target]
X_train,X_test,y_train,y_test=train_test_split(X,y,test_size=.25,random_state=42)
print(X_train.shape, X_test.shape)ML workflow checkpoint 2: Training and test shapes are printed so the split is explicit.
# ML workflow · example 3
from sklearn.dummy import DummyRegressor
from sklearn.metrics import mean_absolute_error
baseline=DummyRegressor(strategy='mean').fit(X_train,y_train)
pred=baseline.predict(X_test)
print('baseline MAE=', round(mean_absolute_error(y_test,pred),3))ML workflow checkpoint 3: The baseline score becomes the minimum standard a more complex model should beat.
# ML workflow · example 4
from sklearn.linear_model import LinearRegression
model=LinearRegression().fit(X_train,y_train)
pred=model.predict(X_test)
print('model MAE=', round(mean_absolute_error(y_test,pred),3))ML workflow checkpoint 4: Compare model MAE with the baseline instead of judging the model in isolation.
# ML workflow · example 5
errors=(y_test-pred).rename('residual')
print(errors.describe().round(3))
print('largest absolute errors:')
print(errors.abs().sort_values(ascending=False).head())ML workflow checkpoint 5: Residual statistics and the largest errors reveal cases worth investigating.
Inspect errors instead of chasing one score
Training performance alone is not evidence of generalization; evaluation data must remain separate from model fitting and tuning.
Error analysis should identify where the model fails so the next iteration targets a real weakness instead of random hyperparameter changes.
Self-check with answers
For ML workflow, what result should you be able to explain without the code?
Machine learning learns patterns from examples, but the quality of labels, split strategy and evaluation often matters more than model complexity.
Which assumption matters most in this ML workflow exercise?
Training performance alone is not evidence of generalization; evaluation data must remain separate from model fitting and tuning.
What failure mode would make this ML workflow result untrustworthy?
Error analysis should identify where the model fails so the next iteration targets a real weakness instead of random hyperparameter changes.
Repeat the experiment with a documented change
Try it yourself workspace
Edit this ML workflow starter, copy it, and run it in the downloadable notebook or your local Jupyter environment.
Questions specific to this lesson
What should I establish before tuning Machine Learning?
For ML workflow, begin with a leakage-safe split and a simple baseline so later tuning has a credible reference point.
How do I check overfitting in ML workflow?
For ML workflow, compare training and validation behavior, inspect held-out errors, and repeat the evaluation with a stable split or cross-validation when appropriate.
Which metric should I report for ML workflow?
Error analysis should identify where the model fails so the next iteration targets a real weakness instead of random hyperparameter changes.
How should I explain ML workflow in a portfolio or interview?
When presenting ML workflow, show the problem definition, baseline, split strategy, metric, one error-analysis finding and one limitation.
Primary references and further reading
For ML workflow, use these primary references when exact syntax, defaults or formal definitions matter beyond the examples on this page.
