Gradient Boosting Regression for Estimating Mackerel Biomass from Acoustic Data and Water Depth

Project info

• Field of studyFisheries and Marine Science
• LevelDoctoral dissertation (S3)
• Method typeRegression
• Estimated build time± 7 working days
• Toolsscikit-learn GradientBoostingRegressor, staged_predict, mean_squared_error
• PriceUS$31 - US$194

Research goal

Model fish biomass from acoustic backscatter and depth using Gradient Boosting Regression and evaluate the model performance.

Variables

Example result

• 0.595R2 (test)
• 10.07RMSE (test)
• 8.167MAE (test)
• 70Best iteration
• 0.08Learning rate
• 0.541Top importance

Gradient boosting reaches a test R2 of 0.595 for Fish biomass (RMSE 10.070); the test error is lowest at iteration 70.

Computed from illustrative synthetic data, not from a real study.

About Gradient Boosting Regression

Gradient boosting regression builds trees sequentially, each correcting the errors of the previous ones. It is used when high predictive accuracy on tabular data is needed and there is time to tune hyperparameters.

Steps

1. 1Split into training and test sets
2. 2Set a small learning_rate and enough trees
3. 3Limit tree depth to avoid overfitting
4. 4Monitor train and test error per iteration
5. 5Evaluate R2 and RMSE on the test set
6. 6Compute feature importance

Assumptions to check

• Enough data for boosting to be stable
• Learning rate, depth, and tree count are tuned on validation, not the test set
• Extreme outliers can affect squared loss

Reading and reporting the results

Report the final hyperparameters, test R2 and RMSE, and the per-iteration error curve to show the stopping point. Compare with random forest or linear regression as baselines.

Example Python code

A short starting script with this project's variable names. The delivered version is adapted to your data and fully commented.

import numpy as np
import matplotlib.pyplot as plt
from sklearn.ensemble import GradientBoostingRegressor
from sklearn.model_selection import train_test_split
from sklearn.metrics import r2_score, mean_squared_error

rng = np.random.default_rng(3)
n = 500
acoustic_backscatter = rng.uniform(-3, 3, n)
depth = rng.normal(size=n)
fish_biomass = acoustic_backscatter ** 2 + 2 * depth + rng.normal(0, 0.5, n)

X = np.column_stack([acoustic_backscatter, depth])
X_tr, X_te, y_tr, y_te = train_test_split(X, fish_biomass, test_size=0.25, random_state=3)
gb = GradientBoostingRegressor(n_estimators=400, learning_rate=0.05, max_depth=3,
                               subsample=0.8, random_state=3)
gb.fit(X_tr, y_tr)
pred = gb.predict(X_te)
print("Test R2:", round(r2_score(y_te, pred), 3))
print("Test RMSE:", round(mean_squared_error(y_te, pred) ** 0.5, 3))
print("Importance:", gb.feature_importances_.round(3))

tr_err = [mean_squared_error(y_tr, p) for p in gb.staged_predict(X_tr)]
te_err = [mean_squared_error(y_te, p) for p in gb.staged_predict(X_te)]
plt.plot(tr_err, label="Train MSE")
plt.plot(te_err, label="Test MSE")
plt.xlabel("Boosting iteration"); plt.ylabel("MSE"); plt.legend()
plt.show()

Packages and prices

Choose how much help you need. Each higher package includes everything in the lower ones. Prices exclude 11% VAT.

• Payment: 50% down payment before work starts, balance after the deliverables are handed over
• Tax: Prices exclude 11% VAT
• Data and academic integrity: Results are computed from your own research data; you write and stand behind your manuscript, we help with the analysis and your understanding of it
• Revisions: Up to 2 revisions are included in the price (Package 3: 3 rounds)

Frequently asked questions