Common Machine Learning Use Cases and Algorithms

Most machine learning (ML) projects fall into one of six main types of problems.  This list is certainly not exhaustive but covers most of the common ML use cases.  

Recommendation systems 

• Using a person’s purchase history to recommend products they are likely to buy
• Recommend TV shows or movies to users based on viewing history and what similar users have watched.  
• Automatically creating playlists for users based on their music streaming history 

Time Series Forecasting 

• Sales forecasting: predicting demand for (or sales of) a product into the future based on seasonal trends 
• Supply chain demand:  predict demand of products by days/weeks/months to help guide planning 
• Predicting a stock price tomorrow based on the history of that stock and other market attributes 
• Estimate the increase in revenue due to different sales strategies  

Computer Vision 

• Counting the number of daily shoppers in a store and how long they spend in store
• Detecting a brand logo in social media posts 
• Autonomous driving 
• Object classification 
• Allowing shoppers to virtually “try on” clothes 
• Optimizing crop planting in agriculture 

Natural Language Processing 

• Extracting topics and sentiment from social media posts 
• Finding names, addresses, or business names in unstructured text
• Classifying documents into similar groups 
• Predicting the next word in sentence to help write documents or emails 
• Sentiment analysis and topic modeling   

Classification and regression 

• Identifying factors associated with high-selling products 
• Predicting the value of a home based on neighborhood, nearby schools, etc. 
• Classifying email as spam or not 
• Determine which product attributes are associated with high-selling products  

More details about each type of ML problem is described below: 

Recommendation Systems 

Recommendation systems are commonly used by websites and mobile apps to offer products to users (see examples below).  These systems work by leveraging historical information (purchase history, viewing history, etc) about user behavior to recommend new items users might like. 

The algorithms behind these recommendations come in two flavors: collaborative filtering and content-based filtering.  Collaborative filtering assumes that users who have similar preferences in the past will have similar preferences in the future.  For example, if one user has purchased items A, B, C, D in the past, and another user has purchased A, B, and C, then item D could be recommended to the second user, because user one liked item D in the past, and the two users have similar tastes.  

Content-based filtering is similar but uses information about the products themselves. These recommendations are based on a user's likes, dislikes, and product attributes.  For example, if a user has purchased music and computer products in the past, other music and computer products could be recommended to them in the future.  


Examples: 

• Amazon.com recommending new items to buy based on the purchase history of users with similar tastes.  
• Netflix recommending new movies or TV shows based on what you watched (“because you watched…”)  
• Spotify creating a new playlist for users based on music they have listened to in the past

Time Series Forecasting

Time series models aim to predict new data in the future.  The classic example is trying to predict the value of a stock tomorrow based on that stock’s history and other data about the market.  A key differentiator of these models is the time dimension - they use data that is specifically linked to moments in time. 

Traditional methods for these types of problems include linear models like ARIMA (‘auto-regressive integrated moving average’) and it’s seasonal variant SARIMA.  Newer methods have successfully used deep learning for these types of problems like recurrent neural networks (RNN), LSTMs, and DeepAR. 

Examples: 

• Predicting a new value of a stock price 
• Demand/Sales forecasting: predicting demand for (or sales of) a product into the future 
• Weather prediction 

Computer Vision 

Computer vision deals with computers recognizing images or parts of images.  Some examples are identifying people, animals, or products in images, or object tracking.  Facial recognition is one type of computer vision example, but companies also use computer vision to automatically find their logo in vast amounts of pictures, visually optimize product packaging, and much more. 

Deep learning plays a big part in computer vision.  Common approaches for computer vision include convolutional neural networks (CNN), pooling, transfer-learning, and Gaussian mixture models (GMM).  

Examples: 

• Object classification 
• Autonomous cars 

Natural Language Processing 

Natural Language Processing (NLP) is focused on text data.  While most data is structured (think spreadsheet data), text data is unstructured (think tweets or blog posts) and is usually more challenging to work with. Extracting meaningful insights from text is a complicated process, but fortunately the field of NLP has created many methods to tackle these problems.  

Common methods in NLP are deep learning algorithms like LSTMs and transformer models, but also include latent dirichlet analysis (LDA), classification algorithms, and conditional models like the linear CRF. NLP models commonly use preprocessing methods like embeddings to more accurately represent words within text.   

Examples: 

• Entity recognition (e.g., finding addresses in a large blob of text)
• Next word prediction
• Topic modeling
• Sentiment analysis  

Classification and Regression 

Classification aims classifying each data point into one of many groups. Usually each data point can only belong to one classification group. For example, using medical data about a patient to predict whether or not they have a disease.  In this case you are trying to classify the patient into one of two groups (‘having disease’ or ‘not having disease’) using other attributes about the patient (height, weight, medical history, etc.).  Classification can extend to many groups - for example, you could classify the temperature tomorrow as ‘low’, ‘medium’ , or ‘high’, or classify potential voters into ‘democrat’, ‘republican’, ‘independent’, ‘other’. 

Some common methods for classification are logistic regression, support vector machines, random forests, and neural networks. 

Regression deals with prediction data that can take on any value.  For example, predicting the exact temperature for a substance or the price of a new house on the market.  Common methods for regression are linear models, feed-forward neural networks, random forests, and gradient boosting methods like XGBoost.  

Examples: 

• Classifying patients as having a disease or not 
• Predicting the price of a new house on the market