ML POC: Machine Learning for
Minimum Viable Model (MVM)

Cloud Deploy: Machine Learning for Minimum Viable Model (MVM) helps customers develop an initial machine learning model running on TensorFlow and Google Cloud Platform to demonstrate proof-of-concept modeling of a specific business use case. Google will guide customers on the iterative process for developing an initial model that demonstrates the feasibility of a solution addressing a business use case.

Key Activities

Data Exploration Guidance

Analyze available data sources to assess state of data and potential usefulness in applying in an ML model.

  • Analyze data characteristics
  • Assess data quality, cleanliness, potential correlation, and patterns
  • Check for class imbalance
  • Validate hypothesis relative to data

Algorithm Selection

Research modeling strategies to determine appropriate ML selection algorithm to address business problem.

  • Research existing strategies and whitepapers
  • Select known algorithms based on hypothesis, type of features, patterns in data
  • Document decisions related to algorithm usage

Feature Engineering

Create ML model features based on raw data analysis and tests.

  • Use domain knowledge to identify potential features
  • Advise on transformation of raw data into feature recommendations
  • Recommend new features, and remove redundant/duplicate features and highly correlated features

Initial Model Development

Develop an initial ML model using the data to solve the business problem, and iterate.

  • Define the right modeling strategy and choose the right ML algorithms
  • Select data set for training, test set, and validation
  • Develop initial model to prove conceptual potential
  • Determine duration and amount of data for initial experiment
  • Evaluate and visualize model result
  • Recommend corrective actions to improve model
  • Iterate and improve model results
Deliverables
  • Preliminary machine learning model through an iterative process
Scope and Pricing
  • Up to 64 FTE days engagement within a two-month period
  • Additional monthly units of service may be required, depending on customer’s business use case or level of data quality
  • Out of scope examples: building a complete data pipeline, deploying the model into production, converting the model into an API
  • Pricing will be agreed upon by customer and Pluto7 and specified in the SOW