Goto

Collaborating Authors

Predicting marketing performance with Machine Learning

#artificialintelligence

It's difficult for companies to judge how a marketing campaign has performed in the past never mind predicting how one will perform in the future. This becomes increasingly more difficult when multiple types of campaigns are ran over the same period, different products are available or unexpected outside factors influence user behaviour. In this project I will attempt, using multiple machine learning models, to judge how marketing campaigns have performed and predict how they will perform in the future. For the modelling I will be using a synthetic dataset containing information on Starbucks members over a month long period. Due to small size of this dataset there are a number of limitations with the data that I will outline later in this post.


To offer or not to offer? A Starbucks Machine Learning Project

#artificialintelligence

Before we get into the machine learning (which is what you are all here for, I know), I started with preliminary data exploration. This helps identify machine learning questions, validate conclusions drawn from machine learning, and find basic statistical descriptors of the dataset that cannot be identified through machine learning alone. First, I explored the amount spent per transaction. As expected for anyone who frequents a coffee shop, the vast majority of transactions are below $8. It is interesting that there is a large peak around $1–2, perhaps these are those people who just go in and get a cup of Pike Place, with room.


K-Nearest Neighbor classification using python

@machinelearnbot

A number of open-source communities are using python to make available artificial intelligence and machine learning related packages and libraries. In this blog I will use libraries from scikit-learn. Project scikit-learn is a Machine Learning Project in Python. It has a good collection of algorithms for some of the well known data-mining and data analysis jobs such as for Classification, Regression, Clustering, Dimensionality reduction and Model Selection. These algorithms are constructed on a stack of NumPy, SciPy library, and matplotlib.



Using AI Chatbot to order coffee

#artificialintelligence

Starbucks has always been ahead of the curve when it comes to marketing and communicating with their customers, and it has paid off big. The company is well-known by customers and is the go-to coffee shop for many. Millennials are a huge chunk of the customers, and they aren't typically so loyal to one business. So when looking for ways to improve your own customer relationships, it isn't a bad thing to look at what Starbucks is doing. The biggest change and most technology-forward move yet was just announced as the new AI Chatbot programming in the Starbucks app will be released soon.