El big data analytics donde menos te lo esperas - Alex Rayón
1. El Big Data Analytics donde
menos te lo esperas
2. Hola!
Soy Alex Rayón
Director Deusto BigData (www.bigdata.deusto.es)
Me puedes encontrar como @alrayon,
en www.alexrayon.es y alex.rayon@deusto.es
3. BIG DATA
Agosto 24 y 25 | Lima – Perú 2018
ANALYTICS SUMMIT
#BIGDATASUMMIT2018
4. ÍNDICE DE CONTENIDOS
El Big Data Analytics donde menos te lo
esperas
Big Data en el Ocio
#BIGDATASUMMIT2018
Aplicación en la agricultura: Mejora en la
producción de Beterraga
31. AGRICULTURA
¿Tabla de Contenidos
1. Why? Project’s use cases
2. How? Descriptive and predictive models
3. What for? Conclusions for stakeholders
32. AGRICULTURA
¿Tabla de Contenidos
1. Why? Project’s use cases
2. How? Descriptive and predictive models
3. What for? Conclusions for stakeholders
33. AGRICULTURA
1. Why? Project’s use cases - Both business and instrumental
Business focuses Use Cases
1. What variables affect the
performance and quality of beet?
2. A cost analysis to determine how to
optimize the profitability obtained
3. Multivariable characterization of
grower: with what profiles of farmers
do we work?
4. What makes a grower change to a
sowing alternative?
Instrumental Use Cases
5. Organize and centralize data and
information to be able to activate it in
data analysis processes
6. Data Quality: identify information
that we do not have well documented
to date to reinforce and enhance
future campaigns
34. AGRICULTURA
1. Why? Project’s use cases - Data pre-processing tasks
✓ 4 Data sources
✓ 205 Data sets
✓ 16 Field notebooks
✓ 22.249 Grower contracts
✓ 5.231 Variables
✓ 524.491 Observations
37. AGRICULTURA
¿Tabla de Contenidos
1. Why? Project’s use cases
2. How? Descriptive and predictive models
3. What for? Conclusions for stakeholders
38. AGRICULTURA
What happened?
➢ Currently, we are able to analyze in one shot all the grower´s performance in order to implant the best strategy
minimizing errors.
➢ At the same time, we´ll find hidden trends so we will anticipate future problems that may happen .
2.1. Descriptive models
39. AGRICULTURA
What will happen and why happened?
➢ There are variables we can't manage, but we can do it in some others, specially if we know the importance of them.
➢ We'll make a “tailor-made” advice report for each grower. We´ll improve productivity or quality areas of the grower´s
performance
2.1. Descriptive models
40. AGRICULTURA
2.3. Business cases
Business focuses Use Cases
2.3.1. What variables affect the
performance and quality of beet?
2.3.2. A cost analysis to determine
how to optimize the profitability
obtained
2.3.3. Multivariable characterization
of growers: with what profiles of
farmers do we work?
2.3.4. What makes a grower
change to a sowing alternative?
41. AGRICULTURA
1. What variables affect the performance and quality of beet?
2.3.1. Business Use Case: Performance and Quality of Beet
42. AGRICULTURA
1. What variables affect the performance and quality of beet?
2.3.1. Business Use Case: Performance and Quality of Beet
43. AGRICULTURA
2.3.1. Business Use Case: Performance and Quality of Beet
1. What variables affect the performance and quality of beet?
Recommendation:
Search within the variables that can be managed those that improve
the performance and quality of each grower. For example:
number of fungicide treatments in the case of performance.
44. AGRICULTURA
2. A cost analysis to determine how to optimize the profitability obtained
2.3.2. Business Use Case: Cost Analysis
45. AGRICULTURA
2. A cost analysis to determine how to optimize the profitability obtained
2.3.2. Business Use Case: Cost Analysis
46. AGRICULTURA
2.3.2. Business Use Case: Cost Analysis
2. A cost analysis to determine how to optimize the profitability obtained
Recommendation:
Study those growers whose irrigation costs are lower to transfer
knowledge to those growers whose irrigation costs are higher
48. AGRICULTURA
2.3.3. Business Use Case: Growers characterization
3. Multivariable characterization of growers: with what profiles of growers do we work?
49. AGRICULTURA
2.3.3. Business Use Case: Growers characterization
3. Multivariable characterization of growers: with what profiles of growers do we
work?
Recommendation:
Continue to obtain data and new variables to define groups of growers with equal
characterization in order to advise them on common problems and best practices.
51. AGRICULTURA
2.3.4. Business Use Case: Farmers’ churn prevention
4. What makes a farmer change to a sowing alternative?
Recommendation:
Continue contributing new data sources such as news in digital media and
its impact on social networks, keywords ... etc. to look for relationship
patterns.
52. AGRICULTURA
¿Tabla de Contenidos
1. Why? Project’s use cases
2. How? Descriptive and predictive models
3. What for? Conclusions for stakeholders
53. AGRICULTURA
3. What for? Final Conclusions
This company is now prepared to start managing the long distance data race.
This is a never ended way as this is a iterative project. Actually, the more data the
models ingest the more strong the results are.
The success of the project begins at the data collection point. It is necessary to
coordinate this stage in the near future to obtain more detail in the data as well as
more fluency.
The next challenges will be to upload all the data to the private cloud, the data ingest
automation and start to define new business cases.