Building revenue from IoT analytics

Considering costs versus benefits as a way to target the optimal ratio that maximizes profit is important in all industries. The following diagram shows how farmers consider cost and the amount of fertilizer versus the resulting yield and crop prices:

Fertilizer price and crop yields. Source: SMART! Fertilizer management

Data scientists spend most of their time arranging and cleaning data. Making their lives easier through the arrangement of data storage can accelerate your learning and the search for value. A method, which we have dubbed Linked Analytic Datasets (LADs), can help.

Designing a progression plan from experimentation to production for your analytics projects will help keep your environment manageable. Setting up a data retention strategy can allow you to hold on to valuable data while still keeping your costs low. All of this was covered in Chapter 11, Strategies to Organize Data for Analytics.

Maximizing profit means keeping costs low while searching for ways to increase revenue. The economics of cloud analytics was covered in Chapter 12, The Economics of IoT Analytics. Cloud environments are not always a panacea, as your costs can climb quickly if not designed appropriately. But keeping a close eye on your bill, while designing to scale up and down to take advantage of cloud economics, can keep your costs low.

Incorporating business case costs and benefits directly into your machine learning modeling will allow you to optimize the model's output. A method to create cost curves in order to visualize the impact of machine learning models was introduced. We also covered a deep dive into predictive maintenance economics with a hypothetical example.

In short, we covered the following recommendations:

  • Organize data for your data scientists, not your database administrators
  • Plan how an initial concept will grow into a production service
  • Develop a data retention plan that retains potential data value
  • Use the cloud appropriately to keep costs low
  • Build costs and benefits into your machine learning process
  • Predictive maintenance does not always make sense; incorporate costs and benefits to determine when it does
..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.
Reset
18.191.211.66