The human insights missing from big data
Ted Talk
在统计学当中也有关于对数据的分析，而大数据，也是这么回事。收到的数据虽然是真实的，而且有一定量可以来分析绝大部分人的偏好还有行为，同时可以顾及到所有人的真实情况。
然而，大数据并不能代表一切，因为大数据是死的，而且来自于历史，来自于过去。当然，它确实可以有之前说的优点。但是呢，大数据呢，并不能全面代表人们的行为。因为人的行为是变化的，而且也是无常的，并不是一成不变的，更甚者，人会学习新东西，也会接受新东西。就好比，以前没有电灯，没有电话，没有电视这些东西，但是经过时间的演变，人们逐渐适应同时一并适用，并且推广到了全世界，其他的东西亦如此。所以，大数据里头并没有显示人们的心情，人们的追求，还有人们的意向。就好比，大数据并不能遇见到人们对新产品的需求，人们对新东西的接受程度还有人们的生活变数。如果没有进入具体的环境中，单单靠大数据并不能模拟出一个真实的世界来的，所以和真实程度的偏差就会大大增加。所以大数据并不是全部，而且一个辅助工具，我们可以通过大数据得到的结果来接近真实结果。
make the big and right important decision for consulting the oracle of future
- Big data: solve the question by making prediction
- Object: big data doesn't help for right decision because investment big data but low return with 73% of data are non-profitable
- the way for using big data: quantifying in the contained system
- Quantification Bias (quant model): the unconscious belief of valuing the measurable over the immeasurable <-- cannot see something outside of it because the environment is complex and unpredictable, especially the human behavior, which can be merged with new factor and change by learning something
- result: relying on big data alone increases the chances we'll miss something while giving us the illusion we know everything
- quantifying is addictive
- the greater risk to be blind in the unknown and consciously thinking that we know
- using thick data: small data size but include incredible data meaning to complement with the quant model
- Integrate big data + think data == form a complete picture
- asking question after collecting data for using data: what this happening?
- Goal: great data insight the algorithm to analyze the question
