259
个编辑
更改
跳到导航
跳到搜索
第249行：
第249行： − + 第255行：
第255行： − 另一个挑战是确定最适合数据的模型和这些模型的复杂性。这些模型将帮助我们预测随着时间的推移社会将如何演变，并为事物如何运作提供可能的解释。+ − + 第265行：
第265行： − 生成模型帮助我们以一种受控的方式进行广泛的定性分析。爱泼斯坦提出的一个模型是基于主体的模拟，该模型通过识别一组初始的异质实体(主体) ，并根据简单的局部规则观察它们的演化和增长。+ 第273行：
第273行： − 但是这些地方性的规则是什么呢？如何为一组异质的代理识别它们？这些规则的评估和影响提出了一整套新的困难。+ − + 第283行：
第283行： − 集成简单的模型在单个任务中表现更好，形成混合模型是一种可以考虑的方法。这些模型可以提供更好的性能和对数据的理解。然而，当需要提出一个组合的、性能良好的模型时，识别并深入理解这些简单模型之间的相互作用的权衡就出现了。此外，开发工具和应用程序来帮助分析和可视化基于这些混合模型的数据是另一个额外的挑战。+
→Model choice and model complexities： 翻译1
===Model choice and model complexities===
===Model choice and model complexities 模型选择和模型复杂性===
Another challenge is identifying the models that would best fit the data and the complexities of these models. These models would help us predict how societies might evolve over time and provide possible explanations on how things work.<ref>Weisberg, Michael. [http://www.academia.edu/download/30604870/lessmorefinal.pdf When less is more: Tradeoffs and idealization in model building]. Diss. Stanford University, 2003.</ref>
Another challenge is identifying the models that would best fit the data and the complexities of these models. These models would help us predict how societies might evolve over time and provide possible explanations on how things work.<ref>Weisberg, Michael. [http://www.academia.edu/download/30604870/lessmorefinal.pdf When less is more: Tradeoffs and idealization in model building]. Diss. Stanford University, 2003.</ref>
Another challenge is identifying the models that would best fit the data and the complexities of these models. These models would help us predict how societies might evolve over time and provide possible explanations on how things work.
Another challenge is identifying the models that would best fit the data and the complexities of these models. These models would help us predict how societies might evolve over time and provide possible explanations on how things work.
另一个挑战是确定最适合数据的模型，以及最合适的'''模型复杂性 Model Complexities'''。这些模型将帮助我们预测随着时间的推移社会将如何演变，并为事物如何运动和变化提供可能的解释。
====Generative models====
====Generative models 生成模型====
Generative models helps us to perform extensive qualitative analysis in a controlled fashion. A model proposed by Epstein, is the agent-based simulation, which talks about identifying an initial set of heterogeneous entities (agents) and observe their evolution and growth based on simple local rules.<ref>Epstein, Joshua M. [https://www.researchgate.net/profile/Eric_Jones14/publication/283615593_Book_Review_-_Generative_Social_Science_Studies_in_Agent-Based_Computational_Modeling/links/5641398808aebaaea1f70216.pdf Generative social science: Studies in agent-based computational modeling]. Princeton University Press, 2006.</ref>
Generative models helps us to perform extensive qualitative analysis in a controlled fashion. A model proposed by Epstein, is the agent-based simulation, which talks about identifying an initial set of heterogeneous entities (agents) and observe their evolution and growth based on simple local rules.<ref>Epstein, Joshua M. [https://www.researchgate.net/profile/Eric_Jones14/publication/283615593_Book_Review_-_Generative_Social_Science_Studies_in_Agent-Based_Computational_Modeling/links/5641398808aebaaea1f70216.pdf Generative social science: Studies in agent-based computational modeling]. Princeton University Press, 2006.</ref>
Generative models helps us to perform extensive qualitative analysis in a controlled fashion. A model proposed by Epstein, is the agent-based simulation, which talks about identifying an initial set of heterogeneous entities (agents) and observe their evolution and growth based on simple local rules.
Generative models helps us to perform extensive qualitative analysis in a controlled fashion. A model proposed by Epstein, is the agent-based simulation, which talks about identifying an initial set of heterogeneous entities (agents) and observe their evolution and growth based on simple local rules.
'''生成模型 Generative Models'''帮助我们以一种受控的方式进行广泛的定性分析。爱泼斯坦（Epstein）提出的一个基于主体的模型，就是探讨如何识别异质实体(主体)的'''初始集 Initial Set'''，并根据简单的局部规则观察它们的演化和增长。
But what are these local rules? How does one identify them for a set of heterogeneous agents? Evaluation and impact of these rules state a whole new set of difficulties.
But what are these local rules? How does one identify them for a set of heterogeneous agents? Evaluation and impact of these rules state a whole new set of difficulties.
但是这些局部规则是什么呢？如何识别一组异质实体？这些规则的评估和影响都是新出现的困难之处。
====Heterogeneous or ensemble models====
====Heterogeneous or ensemble models 异质模型和组合模型====
Integrating simple models which perform better on individual tasks to form a Hybrid model is an approach that can be looked into{{Citation needed|reason=unsupported claims|date=May 2017}}. These models can offer better performance and understanding of the data. However the trade-off of identifying and having a deep understanding of the interactions between these simple models arises when one needs to come up with one combined, well performing model. Also, coming up with tools and applications to help analyse and visualize the data based on these hybrid models is another added challenge.
Integrating simple models which perform better on individual tasks to form a Hybrid model is an approach that can be looked into{{Citation needed|reason=unsupported claims|date=May 2017}}. These models can offer better performance and understanding of the data. However the trade-off of identifying and having a deep understanding of the interactions between these simple models arises when one needs to come up with one combined, well performing model. Also, coming up with tools and applications to help analyse and visualize the data based on these hybrid models is another added challenge.
Integrating simple models which perform better on individual tasks to form a Hybrid model is an approach that can be looked into. These models can offer better performance and understanding of the data. However the trade-off of identifying and having a deep understanding of the interactions between these simple models arises when one needs to come up with one combined, well performing model. Also, coming up with tools and applications to help analyse and visualize the data based on these hybrid models is another added challenge.
Integrating simple models which perform better on individual tasks to form a Hybrid model is an approach that can be looked into. These models can offer better performance and understanding of the data. However the trade-off of identifying and having a deep understanding of the interactions between these simple models arises when one needs to come up with one combined, well performing model. Also, coming up with tools and applications to help analyse and visualize the data based on these hybrid models is another added challenge.
集成在单个任务中表现良好的简单模型并形成'''混合模型 Hybrid model'''是一种可以继续考虑的方法。这些模型有更好的性能并可以对数据提供更好的理解。然而，当需要提出一个整合的、性能良好的模型时，识别简单模型和深入理解模型之间相互作用的权衡问题就出现了（--[[用户:嘉树|嘉树]]（[[用户讨论:嘉树|讨论]]） 此句好长啊，权衡的两个方面是这两个方面吗？囿于知识局限，不理解这两者的矛盾之处）。此外，另一个挑战是开发工具和应用程序来帮助分析和可视化基于这些混合模型的数据。
==Impact 影响==
==Impact 影响==