Feature Selection with Lasso to Identify Key Indicators of tour package sales in travel agencies

Project info

• Field of studyTourism and Hospitality
• LevelMaster's thesis (S2)
• Method typeDimensionality Reduction
• Estimated build time± 5 working days
• Toolsscikit-learn LassoCV, lasso_path, SelectFromModel
• PriceUS$22 - US$131

Research goal

Select the indicators most influential on tour package sales in travel agencies, including advertising budget and package discount, using the Lasso penalty.

Variables

Example result

• 9Selected / total
• 0.005Best alpha
• 0.946R2 (test)
• 0.254RMSE (test, std.)
• 0.068CV MSE min
• 1True recall

Lasso selects 9 of 10 predictors for tour package sales (alpha 0.005) with a test R2 of 0.946; 100% of the truly relevant predictors were recovered.

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

About Feature Selection with Lasso

Feature selection with Lasso uses an L1 penalty so that coefficients of unimportant features become exactly zero. It suits cases with many predictors where the researcher wants a compact variable subset for the final model.

Steps

1. 1Standardise all features
2. 2Split into training and test sets
3. 3Choose alpha with LassoCV on the training data
4. 4Record features with nonzero coefficients
5. 5Refit using only the selected features and evaluate on the test set
6. 6Plot the coefficient path against alpha

Assumptions to check

• Features must be standardised
• Highly correlated features may be picked arbitrarily by Lasso
• Selection is done on training data only to avoid leakage

Reading and reporting the results

Report the best alpha, the list of selected features, and the error compared with the full model. Note that selection indicates predictive usefulness, not causation.

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.linear_model import LassoCV, lasso_path, LinearRegression
from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import train_test_split
from sklearn.metrics import mean_squared_error

rng = np.random.default_rng(23)
n, p = 200, 20
X = rng.normal(size=(n, p))
advertising_budget = X[:, 0]
package_discount = X[:, 1]
tour_package_sales = 4 * advertising_budget - 3 * package_discount + 2 * X[:, 2] + rng.normal(0, 1, n)

X_tr, X_te, y_tr, y_te = train_test_split(X, tour_package_sales, test_size=0.25, random_state=23)
sc = StandardScaler().fit(X_tr)
Xs_tr, Xs_te = sc.transform(X_tr), sc.transform(X_te)

lasso = LassoCV(cv=5, random_state=23).fit(Xs_tr, y_tr)
sel = np.flatnonzero(lasso.coef_)
print("Best alpha:", round(lasso.alpha_, 4))
print("Selected features:", sel)

rmse_full = mean_squared_error(y_te, LinearRegression().fit(Xs_tr, y_tr).predict(Xs_te)) ** 0.5
rmse_sel = mean_squared_error(y_te, LinearRegression().fit(Xs_tr[:, sel], y_tr).predict(Xs_te[:, sel])) ** 0.5
print("RMSE full:", round(rmse_full, 3), "RMSE selected:", round(rmse_sel, 3))

alphas, coefs, _ = lasso_path(Xs_tr, y_tr)
plt.plot(np.log10(alphas), coefs.T)
plt.axvline(np.log10(lasso.alpha_), color="k", ls="--")
plt.xlabel("log10(alpha)"); plt.ylabel("Coefficient")
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