(2) Pre-and mid-term data analysis project: This is a long and troublesome part. The basic data is processed and cleaned in the early stage, and then the monitoring indicators are designed. The design of indicators is not only mathematical analysis, but also needs to be understood by business demanders. After all, the ultimate goal is to let others use it and improve efficiency, rather than highlighting the height of the model. After all the required data are available, the business model (mathematical model) is established, and the whole modeling process is also a process of repeatedly exploring data. In the case of a certain amount of data, the initial modeling application will definitely have this kind of problem, which is very annoying ... and will be adjusted and optimized at the same time as the later application. Skills: database, SQL, excel, R language, mathematical statistics, data mining, business knowledge.
(3) Part-time product manager: After the business model is completed, there will be index results. Put the data into the database. Then you need to find a developer to help you make a visual site. As a data analyst, I know the logic flow, core algorithm and business application of this project best. Find a developer to help you make a visual site: graph, histogram, pie chart, balabala, so that others can see the overall situation of the indicators at a glance. Skills: logical thinking, process planning, data visualization, certain development knowledge (convenience and development communication), expression ability and expression mode.
(4) Since the formal application of the model and indicators: collecting feedback from business departments, constantly communicating with them by email, and constantly optimizing the model and data table. And give some specific needs (temporary needs) analysis and evaluation reports to the business department. Skills: logical thinking, expressive ability
(5) Personal study: Sometimes you will meet the situation of waiting for others' work progress. For example, you can't work at all until the last batch of data from others comes out. Then go online or read books to learn knowledge. Mathematical statistics and data mining are extensive and profound. How to make good use of them and produce the highest cost performance is a science. It doesn't hurt to know more.
(6) Big data part: "Big data" is not my personal work part, but the work content of the whole team. Specifically, people who know hadoop and spark are responsible for running data on it and writing the final implementation code. The division of labor in our group is probably: data analyst, data engineer, (half product manager). There are three kinds of people, some of whom only specialize. Skill score: there is no specific bonus rule, and the team will get bonus points.