Liked what you saw in this project and wish you could do it too? Come join our cool projects, guided by mentors over 4-6 weeks. This is a simple example of how machine learning can be useful to E-commerce businesses, by developing predictions to help them make strategic decisions. We successfully trained a model to predict whether a customer will make a transaction on an E-commerce site. The most appropriate model for this was found to be the Gradient Boosting Classifier, with a 80/20 split, F1 of 0.696 and AUC of 0.801. PageValues-average value for a web page that a user visited before completing an e-commerce transaction (74.6%)īounceRates-percentage of visitors who enter the site from that page and then leave without triggering any other requests during that session (3.7%) by icon developers Smalllikeart and Nikita Golubev, and accessed at. The most important features in the above model are: Logistic regression models for exploit-or-leave decisions. The AUC for the Gradient Boosting Classifier is 0.801, which is considered a good result. A score closer to 1 means that the model is more accurate.īased on the F1 test, the Gradient Boosting Classifier was the most accurate, with an F1 score of 0.696. The F1 score is a measure of the model’s accuracy. Once the models were all run, model performance was assessed using F1 and AUC. Several models were then used to train the data, including In this case, the score to beat is a low F1 score of 0.171, and a Receiver Operating Characteristic (ROC) Area Under Curve (AUC) of 0.506. It acts as the baseline that a model should perform at. A dummy classifier classifies data using simple rules, and is used as a point of comparison for other classifiers. We first used a dummy classifier to determine a baseline performance. The data was split into training and testing data, using an 80:20 split. All features from the dataset were included in the model.

We want to predict whether a visitor to the e-commerce site will make a transaction. Plotting the features into simple bar charts helps us easily identify some trends. Exploratory Data AnalysisĮric first conducted exploratory data analysis to have a clearer view of the dataset. The dataset also includes operating system, browser, region, visitor type (returning or new visitor), whether the date is a weekend or not, and the month of the year. Revenue: Whether or not a sale has been made. The value of this feature is determined by considering the dynamics of e-commerce such as the duration between the order date and delivery date. Mother’s Day, Valentine's Day) in which the sessions are more likely to be finalized with a transaction. Special Day: Indicates the closeness of the site visiting time to a specific special event (e.g. Page Value: Represents the average value for a web page that a user visited before completing an e-commerce transaction. moving from one page to another.īounce Rate: The percentage of visitors who enter the site from that page and then leave (‘bounce’) without triggering any other requests to the analytics server during that session.Įxit Rate: For all page views to the page, the percentage that were the last in the session. These features are derived from the URL information of the pages visited by the user and updated in real time when a user takes an action, e.g. Page Type/Duration: The number of different types of pages visited by a given visitor and the time spent on each page. The dataset contains one year’s worth of data for the following metrics collected via the e-commerce platform: Train machine learning model to predict whether a visitor browsing the site will make a transaction Perform exploratory data analysis on the datasetĬreate dummy variables for categorical features Although it is less pronounced, some evidence suggests that the type of welfare regime influences preferences toward work motives among government employees.These were the steps taken for Project E-Commerce: Also, the preference for work motives held by citizens is correlated with the type of welfare regime in a nation. The findings demonstrate that the locus of an occupation in government and its focus on a public service activity both are important in shaping preferences for work motives related to public service motivation. Is occupational locus or focus important for public service motivation? Does national context influence public service motivation? To answer both questions, the author examines attitudes toward work motives from national samples in 11 North American and Western European nations using multilevel binary logistic regression analysis. Journal Articles Implications of Occupational Locus and Focus for Public Service Motivation : Attitudes Toward Work Motives across Nations Houston, David J.

Logistic regression flaticon