Tableau Automation Toolbox: Overview The role of a data scientist is wide-ranging; typically a data scientist at a startup will have to understand the entirety of an ETL pipeline (from web application to user reports) and be in communication with every team at the company. Reporting isn’t always the most technologically exciting part of a data scientist’s job but it’s vital to understanding data and empowering all employees to promote data-driven decisions.

Supervised Learning Techniques: Background Supervised learning is a machine learning technique where a model learns from labeled data. Since every data input is paired with a corresponding output label, the model can iteratively test various weights and optimize those weights using the known ‘correct’ prediction. This allows the model to pull out complicated patterns from historical data with the goal of predicting the output for new, unseen data. Contrast this with unsupervised learning where the model uses unlabeled data to find patterns without explicit guidance.

Supervised Learning Techniques: Background Supervised learning is a machine learning technique where a model learns from labeled data. Since every data input is paired with a corresponding output label, the model can iteratively test various weights and optimize those weights using the known ‘correct’ prediction. This allows the model to pull out complicated patterns from historical data with the goal of predicting the output for new, unseen data. Contrast this with unsupervised learning where the model uses unlabeled data to find patterns without explicit guidance.

Image Classification: Background Image classification is a widely used type of machine learning with numerous practical applications ranging from face recognition (security) to medical image analysis (diagnostic) to wildlife monitoring (conservation). Convolutional neural networks (CNNs) are frequently used for image classification, in part because of the easy-to-use and well-supported TensorFlow and PyTorch python packages. Additionally CNNs require very little data pre-processing (feature extraction is automated), making them an ideal choice for image classification problems.

Association Rule Learning: Background Association Rule Learning (also called Market Basket Analysis) is a practical and highly interpertable starting place for implementing the first recommendation algorithm for your business. Association rules, or strong relationships between variables in a dataset, can be mined from historical data using an appropriate algorithm. Those rules can then be leveraged to effectively predict future user behavior. Association rules are commonly applied to assist with marketing decisions such as selecting users for a specific ad campaign, recommending personalized services, or smart product up-selling at checkout.

Recommendation Algorithms: Background Popularized by Amazon in the 1990s, recommendation algorithms have become a staple of business analytics. From shopping add-ons to information sharing via contact the influence of recommendation algorithms is undeniable. Under the hood, recommendation algorithms use both implicit (i.e number of times the item is purchased) and explicit (i.e star ratings) user feedback to determine the strength of association for a recommendation. I want to first define two main types of recommendation algorithms; content-based vs collaborative filtering.