The raw data, before any cleaning
Real data is messy. Here is a raw sample straight from the file. A trained eye spots
problems immediately: inconsistent categories (lowercase m / f),
missing values (blank cells), and impossible outliers (a bodyweight no human has).
Roughly half of real data-science work is finding and fixing exactly this.
Cleaning it up
Every fix is a deliberate choice. Standardize the categories, remove the impossible rows, and handle missing values two different ways. You drop rows whose answer (the total) is missing, because you cannot learn from a missing target. But you impute a missing feature like age by filling the median, so you keep the rest of that row's information.
Explore the clean data
Now build intuition. Filter by sex and equipment, and every chart and number updates live as the relationships appear.
Distribution of totals
Bodyweight vs total
Average total by equipment
Average total by age
Build a predictive model
The payoff: teach a model to predict a lifter's total from bodyweight, age, sex, and equipment. We train on 80% of the data and grade on the 20% it never saw. A simple Linear Regression sets the baseline, and a Random Forest is compared against it. On this real, noisier sample the two perform about the same, a useful reminder that a more complex model does not automatically win.
What the model uses to predict
Predict a total
Move the sliders. The model predicts a total live, right in your browser, running the trained linear model's math on your inputs.