October 2022 Summaries

In this Python-based machine learning tutorial, a binary classification algorithm is implemented using the logistic regression model. The Heart failure dataset from Kaggle is used to predict whether a patient has heart disease or not. Data preprocessing steps include handling anomalies and encoding categorical variables. The final model's performance is evaluated using metrics such as accuracy, precision, and recall.

In this interview, Alex Saccois, Senior Director of DataOps & Analytics at Bluecore, and Adam Whitaker, Go-to-Market Analytics Lead, discuss the company's adoption of Hex for revenue forecasting. They explain how they prioritize projects and measure success, as well as their reasons for choosing Hex over other tools. The team built a revenue forecasting app using Hex to help manage client renewals and optimize performance. They also highlight the benefits of Hex, such as its scalability, replicability, and collaboration features. Additionally, they share how Hex has improved their workflow and productivity by allowing them to create powerful reporting and analysis that is fully shareable and customizable.

Linear Regression is a predictive model used to learn the relationship between input variables and a target variable. The algorithm aims to find the optimal values for the slope (m) and y-intercept (b) of the equation y = mx + b, such that for any value of x, it can accurately predict the value of y. In this tutorial, we applied Linear Regression on house price prediction based on the area of the house. We also demonstrated how to improve the accuracy of the model by using multiple linear regression instead of simple linear regression. The multiple regression model showed a 211.8% increase in performance and a 66.67% decrease in margin of error compared to the simple model.

Hex has released a new integration with the dbt Semantic Layer, which includes a Metrics Cell for easy access to trusted metrics without writing code. Users can specify metrics, dimensions, and time grains to get back a data frame for analysis, visualization, and sharing. Additionally, connecting to the dbt Proxy Server allows users to write dbt-flavored SQL in Hex, using refs, macros, and sources directly in queries. This integration enables more people to participate in analytics workflows without needing to know how to write code, while also ensuring consistency and governance of metrics defined by data teams.

Bluecore is a martech platform focused on helping retailers find customers and keep them for life by quickly unlocking and activating retail data through digital campaigns. The company has a performance-driven model that ties its success to customer outcomes, incentivizing both parties to be more effective. Its data warehouse is BigQuery, with dbt used for transformations and Looker for internal business intelligence (BI). Hex is utilized for development, exploratory analysis, prototyping, and some light data engineering/pipelines. The company's data science organization is a core part of its product, providing predictive models based on retail-specific challenges. Bluecore has made significant investments in improving its product through data, leading to performance leaps compared to competitors.

The text discusses three methods for extracting data from a PostgreSQL database and loading it into a Snowflake warehouse. These methods are using connectors like Fivetran or Stitch, using Python as a wrapper for SQL commands, and using SQL utilities from the command line. Each method has its own advantages and is suitable for different situations. The text provides detailed steps on how to implement each of these methods.

The text discusses the complexity and confusion surrounding data job titles, such as Data Analysts, Data Scientists, and Analytics Engineers. It suggests that there are only two fundamental roles in data work: System Builders and Storytellers. System Builders focus on creating integrated, reliable data sets for others, while Storytellers use data to understand the world and influence decisions. The author recommends using this "System Builder vs. Storyteller" paradigm when mapping career growth instead of focusing on specific titles, as it can lead to more efficient teams and happier employees.

In the evolving landscape of data roles, the myriad of titles such as Data Analyst, Data Scientist, and Analytics Engineer often lead to confusion, but can be distilled into two primary functions: System Builders and Storytellers. System Builders, including roles like Data Engineers and Analytics Engineers, focus on creating reliable data systems that empower others, while Storytellers, such as Analysts and Data Scientists, use data to interpret and influence decision-making processes. The complexity of these roles has grown alongside new data tools and technologies, blurring traditional lines based on technical skills. At Hex, the emphasis is on recognizing the distinct contributions of these roles to enhance productivity and clarity within data teams, proposing a structure where both roles collaborate to leverage their strengths. By aligning career growth with the System Builder vs. Storyteller paradigm, organizations can foster happier, more efficient teams, as seen in Hex's own approach, which promotes cross-pollination between roles to achieve greater overall impact.