docs/data-science/the-data-value-creation-loop.md

The Data Value Creation Loop

The Data Value Creation Loop represents the data journey as it progresses from a business problem to raw data, undergoes cleansing and refinement, is used to train models, and finally finds its application in real-world scenarios. Data assets accrue more value at each stage of the loop accrues as it gets closer to its real-world deployment. created when a variety of different skillsets work together; business stakeholders, data engineers, data scientists, MLOps engindeploymenteers, and application developers

• Business Problem: Identifying the business problem that can be addressed with data science is the critical first step. Example: Reducincustomerer churn rate, predicting token prices, or predicting drought risk.
• Raw Data: This is the unprocessed, untouched data, fresh from the source. This data can be static or dynamic from an API. Example: User profiles, historical prices, or daily temperature.
• Cleaned Data and Feature Vectors: The raw data, now polished and transformed into numerical representations - the feature vectors. Example: the coffee shop sales data, now cleaned and organized, or preprocessed text data transformed into word embeddings.
• Trained Models: Machine learning models that have been trained on feature vectors, learning to decode data's patterns and relationships. Example: a random forest model predicting coffee sales or GPT-3 trained on a vast text corpus.
• Data to Tune Models: Additional data introduced to further refine and enhance model performance. Example: a new batch of sales data for the coffee shop model, or specific domain text data for GPT-3.
• Tuned Models: Models that have been optimized for high performance, robustness, and accuracy. Example: a tuned random forest model forecasting the coffee shop's busiest hours, or a fine-tuned GPT-3 capable of generating expert-level text.
• Model Prediction Inputs: Inputs provided to the models to generate insights. Example: inputting today's date and weather into the sales model, or a text prompt for GPT-3 to generate a blog post.
• Model Prediction Outputs: The models' predictions or insights based on the inputs. Example: the sales model's forecast of a spike in iced coffee sales due to an incoming heatwave, or GPT-3's generated blog post on sustainability in business
• Application: Once the models have been deployed and can generate results, they must be packaged into an application so that they can impact real-world scenarios. Build composable user experiences built around the underlying data and model assets.

Let's framework a potential data value creation loop. An insurance provider wants to offer a drought parametric insurance product for farmers in a particular region. Their risk models indicate if they can accurately predict the value of a particular drought index, they will be able to profitably price their product. They launch a data challenge to curate models to accomplish this task.

Those with a strong skill set in data engineering may focus on the beginning of the value creation loop. They create pipelines to ingest climate and local weather data and aggregate them together. They may also include private data provided by the insurance provider. The data engineers publish an asset on Ocean for a regularly updating dataset for computation only

Data scientists can then conduct feature engineering and build their predictive models on top of the data engineers' assets. They train their models with C2D, and then publish them as compute assets on Ocean Market. The data scientists can leverage Ocean to begin attaching immutable proof of the models' accuracy to their assets and build trust in their models value.

Application developers can now work with the insurance provider to incorporate these models into production. The insurance provider can publish their risk model as a compute asset on Ocean. Application developers build different front-end experiences and businesses on top of the models, focusing on customer acquisition and distribution of the product.

With the product in place, farmers can now improve their lives by tapping a valuable product. Revenue is generated by the insurance product and royalties are automatically distributed to the upstream publishers of the data and model assets. Profits can be reinvested to procure greater data to produce better models and more profitable products, kickstarting sustainable value creation for all stakeholders.

The data value creation loop described above is just one in a vast ocean of opportunity. It can be applied across verticals, like DeFi, real estate, climate, sports and healthcare(just to name a few). Endless problems can be addressed by structuring the business problem context, curating the data, building the models, and deploying the assets on Ocean.

Take a look in the next sections to find a few ideas for how you can kickstart you own data value creation loop.