Unboxing Data Science (Short Verison)

Unboxing Data Science (Short Verison)

If you believe data can power and disrupt businesses this talk is for you. In this talk you will join me into an overall unboxing of data science, how I learned and used it to empower a company I worked with, enabling them to take conscious and informed decisions on how to react and plan ahead on different circumstances. A great chance to learn data science and how to apply it in real life.

98195776df79590269541395c699f816?s=128

João Moura

April 20, 2018
Tweet

Transcript

  1. 2.
  2. 3.
  3. 4.
  4. 5.
  5. 8.
  6. 10.

    How much money will we make? Who were our clients?

    Can we forecast a customer's quality?
  7. 21.

    1 1 FORECAST FORECAST MONTH'S REVENUE MONTH'S REVENUE THE NEXT

    THE NEXT What month it is What season it is Last month's revenue Amount spent on Ads Last year's same month revenue } }Linear Regression
  8. 23.

    1 1 FORECAST FORECAST MONTH'S REVENUE MONTH'S REVENUE THE NEXT

    THE NEXT GRADIENT GRADIENT BOOST BOOST REGRESSOR REGRESSOR
  9. 29.
  10. 31.

    EMAILS EMAILS State Country Alexa Rank Customer Orders 2 2

    CLUSTERIZE CLUSTERIZE BY ITS QUALITY BY ITS QUALITY CUSTOMERS CUSTOMERS
  11. 32.

    2 2 CLUSTERIZE CLUSTERIZE BY ITS QUALITY BY ITS QUALITY

    CUSTOMERS CUSTOMERS * All data is normalized, so it's not the actual number of order nor USD spent
  12. 33.

    2 2 CLUSTERIZE CLUSTERIZE BY ITS QUALITY BY ITS QUALITY

    CUSTOMERS CUSTOMERS * All data is normalized, so it's not the actual number of order nor USD spent
  13. 34.

    2 2 CLUSTERIZE CLUSTERIZE BY ITS QUALITY BY ITS QUALITY

    CUSTOMERS CUSTOMERS * All data is normalized, so it's not the actual number of order nor USD spent
  14. 35.

    2 2 CLUSTERIZE CLUSTERIZE BY ITS QUALITY BY ITS QUALITY

    CUSTOMERS CUSTOMERS * All data is normalized, so it's not the actual number of order nor USD spent
  15. 43.

    70% 70% 3 3 FORECAST FORECAST CUSTOMER CUSTOMER A A

    QUALITY QUALITY CROSS VAL SCORE CROSS VAL SCORE
  16. 44.

    89% 89% 3 3 FORECAST FORECAST CUSTOMER CUSTOMER A A

    QUALITY QUALITY SPLIT TEST SCORE SPLIT TEST SCORE
  17. 45.

    3 3 FORECAST FORECAST CUSTOMER CUSTOMER A A QUALITY QUALITY

    ADA ADABOOST BOOSTCLASSIFIER CLASSIFIER ensembling
  18. 48.

    94% 94% 3 3 FORECAST FORECAST CUSTOMER CUSTOMER A A

    QUALITY QUALITY SPLIT TEST SCORE SPLIT TEST SCORE
  19. 49.
  20. 50.