DATA SCIENCE
Over the past five years, companies have invested billions to get the most-talented data scientists to set up shop, amass zettabytes of material, and run it through their deduction machines to find signals in the unfathomable volume of noise. Data has begun to change our relationship to fields as varied as language translation, retail, health care, sports, etc. Wang&Company’s data science services consists of two components – technical and business, from identifying business needs through use case scaling and rollout, we translate data insights into business value.
…the ability to take data — to be able to understand it, to process it, to extract value from it, to visualize it, to communicate it — that’s going to be a hugely important skill in the next decades. (Hal Varian)
The technical component of our data science services consists of five stages, include Capture, Maintain, Process, Analyze and Communicate. Capture means data acquisition, data entry, signal reception, and data extraction; Maintain means data warehousing, data cleansing, data staging, data processing, and data architecture; Process means data mining, clustering and classification, data modeling, and data summarization; Analyze means exploratory and confirmatory, predictive analysis, regression, text mining, and qualitative analysis; Communicate means data reporting, data visualization, and business intelligence.
Technical components of data, analytics (algorithms and technical talent), and IT.
Business components of people (nontechnical talent) and processes.
As data generation and collection grow in volume, data relevance will become more important, since it is impractical and difficult to collect and save every bit of the terabytes or even petabytes of data that were generated every second. To ensure that only relevant data is captured, we will work with our clients to define certain requirements (for data capture) based on business use cases. Our experience showed that using a “hypothesis-driven” and “use case backward” approach when generating and collecting data often delivers better results than capturing all of the available data.
Another important aspect when capturing data is carefully organizing data into multiple logical layers and employing a logic by which to stack these layers instead of taking all relevant data into the analytics layer all at once. We believe that well-prepared data can help generate more meaningful results.
The information and data within your business is a valuable business asset, it can be the key to growth and success. We know the security of your data is a priority within your business, so we take it as seriously as you do. We will follow your business data storage (and/or data security) policies and procedures at all times. (Physical controls, Technical controls, and Administrative controls)
Feature engineering is to create new input features from the existing features. This is one of the most valuable tasks we can do to improve model performance. We will use domain knowledge to isolate, combine, group, and remove features to achieve the goal. Many times, replacing categorical features with dummy variables is necessary since most of data science algorithms are not designed to handle text values.
When extracting insights using machine learning, our experience has proven multiple times that, the combination of human hypothesis-driven input and new surprising patterns that machines reveal is the winning combination. Creating new features just helps the machine find patterns better and also helps humans better describe and act on these patterns. To identify patterns hide in the data, we perform descriptive analytics, predictive analytics, and prescriptive analytics to reach our goal.
- Feature Engineering
- Data Cleaning
- Algorithm Selection
- Analysis
- Other
Once we have extracted important insights from models, it is important to figure out how to turn these insights into action in order to generate business impact. “Who” and “how” both play the key role at this stage. First, hiring the right people who has right domain knowledge is critical; second, even once it becomes clear what action needs to be taken, success can depend on how that action is taken.
In addition, there are a few other common structural challenges that may block businesses from achieving maximum business impact from data. One of the most critical ones is the gap between data science and business execution – a lack of understanding from the business side of what is possible and from the data science side of what is actually needed. But don’t worry, it is our responsibility to help you tackle these challenges to achieve your business goals.