AI for Table Analysis

How do we use AI to extract useful information from tables of data?

Tables are how we organize structured data about our world. They contain information about scientific observations, business transactions, physical entities, and more.

As a result, tables are the most common form of structured data we encounter in information technology.

For the first installment of this multi-part series on The AI Developer’s Toolkit, I’ll introduce you to the three most popular AI tools for tabular data analysis.

Classification

Classification allows us to automatically assign a row of data to one or more categories. It answers the question: “what kind of data is this?” Or “how should we classify this row of data?”

For example, we could use a customer’s data to approve or reject a credit-card application. We provide the classification model with each customer’s financial data as input. Then the model produces a label of “Approve” or “Reject” as output, based on their likelihood of paying their bills on time.

Classification is useful any time we need to automatically assign a row of tabular data into one or more labeled categories. For example:

• automating the approval or rejection of bank loans
• detecting potential fraud during credit-card transactions
• predicting which customers are about to leave us for our competition

Regression

Regression allows us to predict a numerical value based on a row of data. It answers the question: “how much or how many should I assign to this row of data?”

For example, we can use regression to predict the sale price of a house based on its various features. We provide the regression model with a row of data containing all of the features of a house as input. Then the model produces a prediction of the sale price of the house and a confidence score as its output.

Regression is useful anytime we need to estimate a numerical quantity from a row of data. For example:

• forecasting sales for the next day, month, or year
• estimating the potential revenue from each of our customers
• predicting the likelihood of patient survival during medical procedures

Recommendation

Recommendation allows us to suggest products, services, or content to users. It answers the question: “what other things are associated with this thing?”

For example, we can use recommendation to suggest product pairings for grocery-store purchases. We provide the recommendation model with a list of purchases, like chips, milk, and cheese as input. Then the model produces a list of recommendations like salsa, cookies, and crackers as output.

Recommendation is probably the most well-known type of AI tool in the business world. It’s useful any time you want to recommend something that is associated with something else. For example:

• upselling to our clients by recommending new products
• recommending online videos based on viewing history
• suggesting new friends on social media

Other Tools

Beyond these three most common tools, there are also a variety of other tools available for tabular data analysis. For example:

• Clustering – groups rows of data together based on their similarity to one another
• Association – determines the strength of the relationships between connected rows of data
• Anomaly detection – determines which data are atypical (or abnormal) compared to the other data

As we can see, tabular data analysis allows us to extract useful information from tables of data.


If you’d like to learn how to use all of the tools listed above, please watch my online course: The AI Developer’s Toolkit.

The future belongs who those who invest in AI today. Don’t get left behind!

Start Now!