Data science project

As explained in Data Without Labels

A structured project that transforms data into analyses, models, reports, or insights for a defined business problem.

A data science project transforms raw data through coordinated stages into analyses, models, and business decisions. It begins with a concise, clear, measurable, achievable business problem. section 1.4

A practical workflow includes:

• Define the business problem and measurable target. section 1.4
• Discover, access, and organize relevant data sources and tables. section 1.4
• Preprocess data by addressing null values, outliers, duplicates, and junk values. section 1.4
• Explore the data, produce initial reports and insights, and discuss findings with stakeholders. section 1.4
• Test multiple modeling solution versions after the data is prepared. section 1.4

Turning vague goals into measurable problems

Replace “increase production” with a target such as optimizing cost heads A, B, C, and D to reduce cost by 1.2% in six months. For defects, identify significant factors X, Y, and Z to reduce defect percentage by 1.8% in three months. section 1.4

Figure 1.9 A data science project is like any other project, having stages and deadlines, dependencies, and processes.

Use the lifecycle diagram to navigate the project’s stages, deadlines, dependencies, and processes. section 1.4

How Data Without Labels covers it

How deep it goes

Traces broad project stages and details cleaning tasks, but stops short of explaining modeling algorithms or implementation internals.

Examples

Uses customer, purchase, social-media, and portfolio datasets plus measurable cost and defect-reduction goals.

What it assumes

Assumes a defined organizational problem and access to data, while establishing the workflow from discovery through modeling.

What you can do after

Apply a staged workflow for converting a defined business problem and raw data into tested modeling solutions.

How it connects

Connects business problem definition, data discovery, data preprocessing, exploratory data analysis, modeling, and deployment.

liveBookmanning.com

Read Data Without Labels on liveBook Get the book at manning.com

Related concepts

• data explorationhas part · Data exploration is the systematic examination and summarization of data to understand its structure, quality, distributions, relationships, patterns, and anomalies.
• preprocessinghas part · Preprocessing is the preparation or transformation of data, content, or other inputs into a form suitable for a subsequent computational operation.
• data discoveryhas part · Data discovery is the process of locating, examining, characterizing, and evaluating data sources, assets, contents, relationships, and metadata.

This explainer was generated by Manning's knowledge graph from the text of Data Without Labels. The section links point to the parts of the book it draws on.