Next Steps

Where to go from here to deepen your data science and ML skills.

You have seen the whole stack: NumPy, pandas, visualisation, preparation, classic ML, and a taste of deep learning. Here is how to grow.

Practise

  • Enter beginner Kaggle competitions (Titanic, House Prices).
  • Rebuild these examples on your own datasets.

Go deeper

  • Learn pipelines and GridSearchCV for tuning.
  • Explore gradient boosting (XGBoost, LightGBM).
  • Study deep learning with PyTorch or TensorFlow.

The best way to learn is to build. Pick a dataset you care about and go.

Example

Example · python
# A production-style pipeline bundles preprocessing + model
from sklearn.pipeline import make_pipeline
from sklearn.preprocessing import StandardScaler
from sklearn.linear_model import LogisticRegression

pipe = make_pipeline(StandardScaler(), LogisticRegression())
# pipe.fit(X_train, y_train); pipe.predict(X_test)
print('One object handles scaling AND the model together')

When to use it

  • A junior data scientist follows a structured progression: practice Kaggle competitions to apply sklearn skills, then take a fast.ai course to add deep learning.
  • An ML engineer sets up a GitHub portfolio of three end-to-end notebooks (regression, classification, clustering) to demonstrate skills to future employers.
  • A researcher identifies specific gaps (NLP, time series, MLOps) after completing the fundamentals and targets one library or course per gap.

More examples

Explore a Kaggle dataset locally

Downloads a public Kaggle dataset using the Kaggle CLI and immediately performs an initial audit using pandas, a first practical next-step exercise.

Example · python
# Install Kaggle CLI and download a competition dataset
# pip install kaggle
# kaggle competitions download -c titanic -p ./data

import pandas as pd
df = pd.read_csv('./data/train.csv')
print(df.shape)
print(df.isnull().sum())
print(df.describe())

Submit to Kaggle from a notebook

Trains a Gradient Boosting model on the Titanic dataset and saves a submission CSV, walking through the complete Kaggle competition workflow.

Example · python
import pandas as pd
from sklearn.ensemble import GradientBoostingClassifier
from sklearn.preprocessing import LabelEncoder

train = pd.read_csv('./data/train.csv').dropna(subset=['Survived'])
test  = pd.read_csv('./data/test.csv')

features = ['Pclass', 'Age', 'SibSp', 'Parch', 'Fare']
X = train[features].fillna(train[features].median())
y = train['Survived']

model = GradientBoostingClassifier(n_estimators=200).fit(X, y)
X_te = test[features].fillna(train[features].median())

pd.DataFrame({'PassengerId': test.PassengerId, 'Survived': model.predict(X_te)}).to_csv('submission.csv', index=False)
print('submission.csv saved')

Roadmap as executable checklist

Prints a structured learning roadmap as a checklist of skills, marking each as done or to-do, providing a clear, personalised path for continuing ML education.

Example · python
roadmap = [
    ('numpy & pandas',  True,  'core data manipulation'),
    ('matplotlib',      True,  'visualisation'),
    ('scikit-learn',    True,  'classical ML'),
    ('Keras / PyTorch', False, 'deep learning'),
    ('NLP (spaCy)',     False, 'text models'),
    ('MLOps (MLflow)',  False, 'experiment tracking'),
    ('Cloud deploy',    False, 'production serving'),
]

for skill, done, desc in roadmap:
    status = 'DONE' if done else 'TODO'
    print(f'[{status}] {skill}: {desc}')

Related

pyai project workflowpyai ml sklearn api

Discussion

  • Be the first to comment on this lesson.