jimaworks Tech Consulting
Advising/Implementation/Analysis for Data Mining, GeoAnalytics, and Computer Vision
jimaworks founder Jim Antonisse spend 35 years in Research and Development for the US Government. That experience drove into him an abiding committment to concrete, standards-based deliveries of advanced technologies in small, component-like packages
It is in that spirit that jimaworks Consulting offers the following structured approach to its analytics. The jimaworks approach is built directly on the industry-standard Data Mining process standard CRISP-DM, deepened with specific tools and tasking, and expanded to include Computer Vision and Geo-Analytics as well as Data Mining.
The key innovations that jimaworks adds to this process are:
- formalization of the standard stages of CRISP-DM" in terms of concrete deliverables
- per-stage pricing of the application of jimaworks CRISP analytics
- per-stage deliveries to ensure satisfactory progress
- formalized next-stage requirements to ensure well-founded workflow and results
- "off-ramps" at each stage for customer control of pause/stop of project
The jimaworks' instantiation of CRISP
On Ramp = General business goals, a sense of Data Mining risks, costs, benefits.
- Gather background information
- Assess the business and data situation
- Determine the data and information goals
- Develop component subtasks and milestones
Deliverable/Off Ramp Product: Project Plan, including Task Data goals, risks, costs, benefits. $1000
On Ramp = Understanding of the business implications and execution parameters of the job, Project Plan.
- Collect initial data
- Describe the data
- Explore data characteristics
- Verify data quality
Deliverable/Off Ramp Product: Data Prospects Report, including data risks, data costs, data benefits $1000
On Ramp: Sepcific data source, type, quality, and quantity goals.
- Select a focus data and feature set for this phase
- Clean the data
- Extend the data
- Format the data
Deliverable/Off Ramp Product: Data Quality/Quantity Report, including correctness, completeness, coherence, and data risk/cost/benefit assessment $1000
On Ramp: Sufficient quantities of adequately relevant, clean, correct data sets in model-ingestible formats.
- Select modeling technique for this phase
- Build the model
- Assess the model
Deliverable/Off Ramp Product: Model Report, including parameter/hyperparameter optimization results $1000
On Ramp: Executable model and suitable test sets.
- Evaluate the results
- Review the process
- Determine next steps
Deliverable/Off Ramp Product: Model Performance and Evaluation Report, including next-step recommendations $1000
On Ramp: Executable model, its evaluation, and suitable cost predictors.
- Plan for sustained use/application
- Plan monitoring and maintenance
- Conduct final project review