推理 & 模型 Inference and Models

Grade 2 ( 5-7岁 ) 要掌握的程度

Data can be used to make inferences or predictions about the world. Inferences, statements about something that cannot be readily observed, are often based on observed data. Predictions, statements about future events, are based on patterns in data and can be made by looking at data visualizations, such as charts and graphs.

Observations of people’s clothing (jackets and coats) can be used to make an inference about the weather (it is cold outside). Patterns in past data can be identified and extrapolated to make predictions. For example, a person’s lunch menu selection can be predicted by using data on past lunch selections.

Crosscutting Concept: Abstraction Connection Within Framework: K–2.Impacts of Computing.Culture

可以使用数据来对世界进行推断和预测:

• 推断: 对不易观察到的事物进行描述, 通常是基于观察到的数据
• 预测: 对未来的预测, 通常是基于从数据中发现特定的模式, 可以通过观察数据的可视化(比如图表和图形)来发现这些模式

比如说, 通过观察人们穿的衣服可以用来推断天气状况. 比如说, 看到人们穿着夹克和大衣, 可以判断外面天气很冷.

用于预测的数据模式, 可以从过去数据中识别和推断出来. 比如, 一个人午餐选择吃什么, 可以通过他过去的午餐选择的数据来进行预测.

Grade 5 ( 8-11岁 ) 要掌握的程度

The accuracy of inferences and predictions is related to how realistically data is represented. Many factors influence the accuracy of inferences and predictions, such as the amount and relevance of data collected.

People use data to highlight or propose cause-and-effect relationships and predict outcomes. Basing inferences or predictions on data does not guarantee their accuracy; the data must be relevant and of sufficient quantity. An example of irrelevance is using eye color data when inferring someone’s age. An example of insufficient quantity is predicting the outcome of an election by polling only a few people.

Crosscutting Concept: System Relationships

许多因素影响推断和预测的准确性, 比如收集数据的数量、数据的真实性、数据的相关性.

我们可以使用数据来强调某个方面, 或者使用数据来提出因果关系并预测结果. 但是要注意, 基于数据的推断或预测并不一定是准确的. 如果数据是不相关的, 或者数据量不够, 都会降低推断和预测的准确性. 比如说, 使用眼睛的颜色数据来推断这个人的年龄, 显然眼睛颜色的数据和年龄是不相关的, 由此带来的年龄推断肯定是不准确的. 再比如说, 只通过对少数人的调查统计来预测选举结果肯定是不准的, 因为统计数量不足.

Grade 8 ( 11-14岁 ) 要掌握的程度

Computer models can be used to simulate events, examine theories and inferences, or make predictions with either few or millions of data points. Computer models are abstractions that represent phenomena and use data and algorithms to emphasize key features and relationships within a system. As more data is automatically collected, models can be refined.

Very large data sets require a model for analysis; they are too large to be analyzed by examining all of the records. While individual users are online, shopping websites and online advertisements use personal data they generate, compared to millions of other users, to predict what they would like and make recommendations. A video-streaming website may recommend videos based on models generated from other users and based upon their personal habits and preferences. The data that is collected about an individual and potential inferences made from that data can have implications for privacy.

Crosscutting Concepts: Privacy and Security; Abstraction Connections Within Framework: 6–8.Algorithms and Programming.Algorithms; 6–8.Impacts of Computing.Culture

建立计算模型非常有用, 它可以用来模拟事件、检查理论和推论、帮助我们使用较少的数据来进行预测.

计算模型是对现实的抽象, 它是利用数据和算法, 来抽离出系统中的关键特征和主要关系. 好消息是, 通过收集更多的数据对模型进行验证, 可以帮助我们完善模型.

大型的数据集需要使用计算模型进行分析. 因为数据量过于庞大, 无法事无巨细地检查所有数据来进行分析. 比如说, 当用户上网时, 购物网站和在线广告会使用他们产生的个人数据, 和其他数百万用户生成的计算模型进行比较, 以此来预测用户会喜欢什么, 并进行推荐. 再比如说, 视频网站利用其他用户的数据搭建计算模型, 然后结合这个模型和当前用户的个人习惯和喜好来推荐视频.

需要注意的是, 收集的个人数据以及从这些数据中做出的推断预测, 可能会侵犯用户的隐私.

Grade 12 ( 14-18岁 ) 要掌握的程度

The accuracy of predictions or inferences depends upon the limitations of the computer model and the data the model is built upon. The amount, quality, and diversity of data and the features chosen can affect the quality of a model and ability to understand a system. Predictions or inferences are tested to validate models.

Large data sets are used to make models used for inference or predictions, such as forecasting earthquakes, traffic patterns, or the results of car crashes. Larger quantities and higher quality of collected data will tend to improve the accuracy of models. For example, using data from 1,000 car crashes would generally yield a more accurate model than using data from 100 crashes. Additionally, car crashes provide a wide variety of data points, such as impact speed, car make and model, and passenger type, and this data is useful in the development of injury prevention measures.

Crosscutting Concepts: Abstraction; Privacy and Security

推断和预测的准确性, 取决于计算模型的局限性和模型所基于的数据. 数据的数量、质量、多样性以及选择的特征, 会影响模型的质量. 反过来, 推断和预测的结果也可以用来验证和修正模型.

大型的数据集被用来搭建用于推断和预测的计算模型, 比如用来预测地震、模拟交通、预测车祸.

较大数量和较高质量的数据将提高计算模型的准确性. 比如说, 基于 1000 次车祸数据产生的模型, 通常会比基于 100 次车祸数据产生的模型更准确. 另外, 车祸这个事件其实提供了各种各样的数据, 比如撞击速度、汽车品牌、汽车型号、乘客类型等等, 这些数据在制定汽车安全预防措施时非常有用.