estimation

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Estimation

估算

When to Use This Skill

何时使用本技能

Use this skill when:

Estimation tasks - Working on estimation techniques including analogous, parametric, three-point, and expert judgment methods. provides effort, cost, duration, and complexity estimates for projects, features, and tasks
Planning or design - Need guidance on Estimation approaches
Best practices - Want to follow established patterns and standards

当你遇到以下场景时可使用本技能：

估算任务：处理涉及类比法、参数法、三点法和专家判断法的估算工作，为项目、功能和任务输出工作量、成本、周期和复杂度估算结果
规划或设计阶段：需要获取估算方法相关指导
遵循最佳实践：希望遵循成熟的模式和标准开展工作

Overview

概述

Systematically estimate effort, cost, duration, and complexity using proven estimation techniques. Supports analogous, parametric, three-point (PERT), and expert judgment methods for projects, features, and tasks.

使用经过验证的估算技术，系统地估算工作量、成本、周期和复杂度。支持对项目、功能和任务使用类比法、参数法、三点法（PERT）和专家判断法。

What is Estimation?

什么是估算？

Estimation is the process of forecasting the resources, time, and cost required to complete work. Good estimation balances:

Accuracy: How close to actual results
Precision: Consistency of estimates
Speed: Time to produce estimates
Communication: Shared understanding of uncertainty

估算是预测完成工作所需资源、时间和成本的过程。优质的估算需要平衡以下维度：

准确性：估算结果与实际值的接近程度
精确性：多次估算结果的一致性
速度：生成估算结果花费的时间
沟通性：各方对估算不确定性的共识程度

Estimation vs Commitment

估算 vs 承诺

Concept	Definition	Use
Estimate	Best guess given current knowledge	Planning, forecasting
Commitment	Promise to deliver by date/cost	Contracts, deadlines
Target	Desired outcome to aim for	Goals, objectives

Key principle: Estimates are ranges with uncertainty, not single-point guarantees.

概念	定义	用途
估算	基于当前信息得出的最优判断	规划、预测
承诺	按约定日期/成本交付的承诺	合同、截止时间
目标	期望达成的结果	目的、指标

核心原则：估算是带不确定性的区间值，而非单点保证。

Estimation Techniques

估算技术

Analogous Estimation (Top-Down)

类比估算（自上而下）

Estimate based on similar past work:

Step	Action
1	Identify similar completed project/feature
2	Retrieve actual effort/cost/duration
3	Adjust for differences (complexity, team, technology)
4	Apply adjustment factor

Formula:

text

New Estimate = Historical Actual × Adjustment Factor

Example:
Similar project took 200 hours
New project is ~20% more complex
Estimate = 200 × 1.20 = 240 hours

When to Use: Early phases, limited detail, experienced teams with historical data

Accuracy: +/- 25-50% (improves with good historical data)

基于类似的过往工作进行估算：

步骤	操作
1	找到相似的已完成项目/功能
2	获取该项目的实际工作量/成本/周期
3	根据差异（复杂度、团队、技术栈）调整
4	应用调整系数计算结果

公式：

text

New Estimate = Historical Actual × Adjustment Factor

Example:
Similar project took 200 hours
New project is ~20% more complex
Estimate = 200 × 1.20 = 240 hours

适用场景： 项目早期阶段、细节信息有限、团队有丰富历史数据积累

准确率： +/- 25-50%（历史数据质量越高准确率越高）

Parametric Estimation

参数估算

Estimate using statistical relationships:

Element	Description
Unit of Work	Measurable element (screen, API, table)
Productivity Rate	Effort per unit from historical data
Quantity	Number of units to produce

Formula:

text

Estimate = Quantity × Productivity Rate

Example:
10 API endpoints × 16 hours/endpoint = 160 hours

Common Productivity Metrics:

Work Type	Metric	Typical Range
UI screens	Hours/screen	8-40 hours
API endpoints	Hours/endpoint	4-24 hours
Database tables	Hours/table	4-16 hours
Test cases	Hours/test	0.5-4 hours
Documentation pages	Hours/page	2-8 hours

When to Use: Repeatable work, good historical data, similar technology

Accuracy: +/- 15-25% (with calibrated rates)

使用统计关系进行估算：

元素	描述
工作单元	可量化的工作项（页面、API、数据库表）
生产率	基于历史数据得出的单个工作单元所需工作量
数量	需要完成的工作单元总数

公式：

text

Estimate = Quantity × Productivity Rate

Example:
10 API endpoints × 16 hours/endpoint = 160 hours

常用生产率指标：

工作类型	指标	典型范围
UI页面	小时/页	8-40小时
API接口	小时/个	4-24小时
数据库表	小时/张	4-16小时
测试用例	小时/个	0.5-4小时
文档页面	小时/页	2-8小时

适用场景： 可复用的重复工作、历史数据完善、技术栈相似度高

准确率： +/- 15-25%（生产率校准后可提升）

Three-Point Estimation (PERT)

三点估算（PERT）

Estimate using optimistic, most likely, and pessimistic values:

Value	Symbol	Definition
Optimistic	O	Best case, everything goes right
Most Likely	M	Most probable outcome
Pessimistic	P	Worst case, problems occur

PERT Formula (Weighted Average):

text

Expected = (O + 4M + P) / 6
Standard Deviation = (P - O) / 6

Example:
O = 5 days, M = 8 days, P = 17 days
Expected = (5 + 4×8 + 17) / 6 = 9 days
Std Dev = (17 - 5) / 6 = 2 days

Confidence Intervals:

Confidence	Calculation	Example (E=9, SD=2)
68%	E ± 1 SD	7-11 days
95%	E ± 2 SD	5-13 days
99.7%	E ± 3 SD	3-15 days

When to Use: Uncertain work, new technology, need to communicate risk

Accuracy: Provides explicit uncertainty range

使用乐观值、最可能值、悲观值三个维度进行估算：

值	符号	定义
乐观值	O	最优场景，所有环节都顺利
最可能值	M	概率最高的结果
悲观值	P	最差场景，出现各类问题

PERT公式（加权平均）：

text

Expected = (O + 4M + P) / 6
Standard Deviation = (P - O) / 6

Example:
O = 5 days, M = 8 days, P = 17 days
Expected = (5 + 4×8 + 17) / 6 = 9 days
Std Dev = (17 - 5) / 6 = 2 days

置信区间：

置信度	计算方式	示例（期望值=9，标准差=2）
68%	期望值 ± 1倍标准差	7-11天
95%	期望值 ± 2倍标准差	5-13天
99.7%	期望值 ± 3倍标准差	3-15天

适用场景： 不确定性高的工作、新技术栈、需要沟通风险

准确率： 可输出明确的不确定性区间

Expert Judgment

专家判断

Estimate using collective expert knowledge:

Wideband Delphi Process:

Round	Activity
1	Experts estimate independently
2	Collect and share anonymous estimates
3	Discuss high/low outliers, share rationale
4	Re-estimate independently
5	Repeat until convergence (or average)

Planning Poker (Agile):

Step	Action
1	Present item to estimate
2	Discuss briefly (2-5 minutes)
3	Each team member selects card privately
4	Reveal simultaneously
5	Discuss outliers
6	Re-vote until consensus

Common Scales:

Scale Type	Values
Fibonacci	1, 2, 3, 5, 8, 13, 21, 34, 55, 89
Modified Fibonacci	0, 0.5, 1, 2, 3, 5, 8, 13, 20, 40, 100
T-shirt	XS, S, M, L, XL, XXL
Powers of 2	1, 2, 4, 8, 16, 32, 64

When to Use: Complex work, multiple perspectives needed, team alignment

基于专家集体知识进行估算：

宽带德尔菲流程：

轮次	活动
1	专家独立给出估算结果
2	收集并匿名共享所有估算结果
3	讨论最高/最低异常值，各自说明判断依据
4	专家再次独立给出估算结果
5	重复流程直到结果收敛（或取平均值）

计划扑克（敏捷场景）：

步骤	操作
1	展示待估算的工作项
2	简短讨论（2-5分钟）
3	所有团队成员私下选择对应点数的卡片
4	同时亮出所有卡片
5	讨论异常值的判断依据
6	重新投票直到达成共识

常用估算尺度：

尺度类型	取值
斐波那契数列	1, 2, 3, 5, 8, 13, 21, 34, 55, 89
改良斐波那契数列	0, 0.5, 1, 2, 3, 5, 8, 13, 20, 40, 100
T恤尺寸	XS, S, M, L, XL, XXL
2的幂	1, 2, 4, 8, 16, 32, 64

适用场景： 复杂工作、需要多视角评估、需要团队对齐认知

Function Point Analysis

功能点分析

Estimate based on functional size (for larger systems):

Component	Description	Weight Range
External Inputs	Data entering system	3-6
External Outputs	Data leaving system	4-7
External Inquiries	Read-only queries	3-6
Internal Files	Logical data stores	7-15
External Interfaces	Shared data	5-10

Process:

Count each component type
Classify complexity (low/medium/high)
Apply weights
Calculate unadjusted function points
Apply technical complexity factor

When to Use: Large projects, formal contracts, industry benchmarking

基于功能规模进行估算（适用于大型系统）：

组件	描述	权重范围
外部输入	进入系统的数据	3-6
外部输出	流出系统的数据	4-7
外部查询	只读查询	3-6
内部文件	逻辑数据存储	7-15
外部接口	共享数据	5-10

流程：

统计每类组件的数量
按复杂度分类（低/中/高）
应用对应权重
计算未调整功能点
应用技术复杂度系数

适用场景： 大型项目、正式合同、行业基准对比

Relative Estimation

相对估算

Story Points

故事点

Relative complexity/effort measure:

Points	Relative Size	Example
1	Trivial	Fix typo, config change
2	Simple	Simple bug fix, minor feature
3	Moderate	Standard feature, moderate complexity
5	Complex	Multi-component feature
8	Very complex	Integration work, significant unknowns
13	Epic-sized	Consider breaking down
21+	Too large	Must decompose

Baseline: Pick a well-understood story as reference (e.g., "this is a 3")

Velocity: Story points completed per iteration (used for forecasting)

衡量相对复杂度/工作量的单位：

点数	相对规模	示例
1	极简单	修复错别字、修改配置
2	简单	简单bug修复、小型功能
3	中等	标准功能、中等复杂度
5	复杂	多组件联动的功能
8	非常复杂	集成工作、存在大量未知项
13	史诗级规模	建议拆分
21+	规模过大	必须拆分

基准： 选择一个共识度高的故事作为参考（比如“这个工作项是3点”）

速率： 每个迭代完成的故事点总数（用于后续预测）

T-Shirt Sizing

T恤尺寸估算

Quick relative sizing for roadmap planning:

Size	Effort Range	Duration Range
XS	1-4 hours	< 1 day
S	0.5-2 days	1-2 days
M	2-5 days	3-5 days
L	1-2 weeks	1-2 weeks
XL	2-4 weeks	2-4 weeks
XXL	1-3 months	Too big, decompose

用于路线图规划的快速相对估算方法：

尺寸	工作量范围	周期范围
XS	1-4小时	< 1天
S	0.5-2天	1-2天
M	2-5天	3-5天
L	1-2周	1-2周
XL	2-4周	2-4周
XXL	1-3个月	规模过大，需要拆分

Workflow

工作流程

Phase 1: Prepare

阶段1：准备

Step 1: Clarify Scope

步骤1：明确范围

markdown

undefined

markdown

undefined

Estimation Request

估算请求

Item: [What's being estimated] Requester: [Who needs the estimate] Purpose: [Planning/budgeting/commitment] Deadline: [When estimate is needed] Precision: [ROM/Budget/Definitive]

工作项： [待估算的内容] 请求方： [需要估算结果的人员/团队] 用途： [规划/预算/承诺] 截止时间： [需要交付估算结果的时间] 精度要求： [粗略量级/预算级/确定性级]

Scope Definition

范围定义

In Scope: [What's included]
Out of Scope: [What's excluded]
Assumptions: [Key assumptions]
Constraints: [Known constraints]

undefined

包含范围： [估算包含的内容]
排除范围： [估算不包含的内容]
假设条件： [核心假设]
约束条件： [已知约束]

undefined

Step 2: Select Estimation Technique

步骤2：选择估算技术

Situation	Recommended Technique
Early project phase	Analogous + T-shirt sizing
Detailed requirements	Parametric + Three-point
Agile backlog	Story points + Planning poker
New technology/domain	Expert judgment + Three-point
Contract/budget	Function points + Parametric

场景	推荐技术
项目早期阶段	类比法 + T恤尺寸估算
需求细节明确	参数法 + 三点法
敏捷需求池	故事点 + 计划扑克
新技术/新领域	专家判断 + 三点法
合同/预算场景	功能点 + 参数法

Phase 2: Estimate

阶段2：执行估算

Step 1: Decompose Work

步骤1：拆分工作

Break down into estimable units (half-day to 2-week chunks)

将工作拆分为可估算的单元（粒度控制在半天到2周之间）

Step 2: Apply Technique

步骤2：应用估算技术

Use selected technique(s) from above

使用上文选择的估算技术开展估算

Step 3: Add Contingency

步骤3：添加应急储备

markdown

undefined

markdown

undefined

Contingency Calculation

应急储备计算

Risk Level	Contingency	When to Use
Low	10-15%	Well-understood, experienced team
Medium	20-30%	Some unknowns, new team members
High	40-50%	Significant unknowns, new technology
Very High	75-100%	Research, innovation, first-of-kind

undefined

风险等级	应急储备比例	适用场景
低	10-15%	工作内容清晰、团队经验丰富
中	20-30%	存在部分未知项、有新团队成员
高	40-50%	存在大量未知项、使用新技术
极高	75-100%	研究类、创新类、首次尝试的工作

undefined

Step 4: Validate

步骤4：验证结果

Sanity check against similar work
Review with team
Check for missing items
Verify assumptions

和类似过往工作做合理性校验
和团队一起评审
检查是否有遗漏的工作项
验证假设条件是否成立

Phase 3: Communicate

阶段3：沟通结果

Step 1: Express as Range

步骤1：以区间形式输出

markdown

undefined

markdown

undefined

Estimate Summary

估算汇总

Effort: 160-200 hours (confidence: 80%) Duration: 4-5 weeks (with 2 developers) Cost: $24,000-$30,000

Key Risks:

API integration complexity unknown
Dependency on third-party availability

undefined

工作量： 160-200小时（置信度：80%） 周期： 4-5周（2名开发人员投入） 成本： $24,000-$30,000

核心风险：

API集成复杂度未知
依赖第三方服务的可用性

undefined

Step 2: Document Assumptions

步骤2：记录假设条件

markdown

undefined

markdown

undefined

Estimation Assumptions

估算假设条件

Requirements are stable and complete
Team has 80% availability (20% overhead)
No major technology changes
Dependencies delivered on time
[Additional assumptions...]

If assumptions change, estimate should be revisited.

undefined

需求稳定且完整
团队可用度为80%（20%为 overhead）
无重大技术变更
依赖项按时交付
[其他假设条件...]

如果假设条件发生变化，需要重新评估估算结果。

undefined

Output Formats

输出格式

Narrative Summary

文字总结

markdown

undefined

markdown

undefined

Estimation Summary

估算汇总

Item: [Feature/Project name] Date: [ISO date] Estimator: estimation-analyst Technique: [Technique used]

工作项： [功能/项目名称] 日期： [ISO格式日期] 估算人： estimation-analyst 使用技术： [使用的估算技术]

Estimate

估算结果

Dimension	Low	Expected	High	Confidence
Effort	120h	160h	220h	80%
Duration	4w	5w	7w	80%
Cost	$18K	$24K	$33K	80%

维度	下限	期望值	上限	置信度
工作量	120h	160h	220h	80%
周期	4w	5w	7w	80%
成本	$18K	$24K	$33K	80%

Basis of Estimate

估算依据

Historical Reference: [Similar past work]
Productivity Rate: [If parametric]
Expert Input: [Who contributed]

历史参考： [类似过往工作]
生产率： [参数法适用]
专家输入： [参与估算的人员]

Assumptions

假设条件

[Assumption 1]
[Assumption 2]

[假设条件1]
[假设条件2]

Risks Affecting Estimate

影响估算的风险

Risk	Impact on Estimate
[Risk 1]	+20% if occurs
[Risk 2]	+15% if occurs

风险	对估算的影响
[风险1]	发生的话增加20%工作量
[风险2]	发生的话增加15%工作量

Recommendations

建议

[Next steps for refining estimate]
[When to re-estimate]

undefined

[细化估算的下一步动作]
[重新估算的触发条件]

undefined

Structured Data (YAML)

结构化数据（YAML）

yaml

estimation:
  version: "1.0"
  date: "2025-01-15"
  item: "User Dashboard Feature"
  estimator: "estimation-analyst"
  technique: "three_point"

  scope:
    description: "Interactive user dashboard with analytics"
    in_scope:
      - "Dashboard UI components"
      - "Data visualization"
      - "User preferences"
    out_of_scope:
      - "Backend analytics engine"
      - "Real-time updates"
    assumptions:
      - "API endpoints available"
      - "Design mockups complete"

  estimates:
    effort:
      optimistic: 120
      most_likely: 160
      pessimistic: 240
      expected: 166
      unit: "hours"
      std_deviation: 20
    duration:
      optimistic: 3
      most_likely: 4
      pessimistic: 6
      expected: 4.2
      unit: "weeks"
    cost:
      expected: 24000
      range_low: 18000
      range_high: 36000
      currency: "USD"

  confidence: 0.80
  contingency: 0.20

  breakdown:
    - component: "UI Components"
      effort: 60
      technique: "parametric"
      rate: "15h/component"
      quantity: 4
    - component: "Data Integration"
      effort: 40
      technique: "analogous"
      reference: "PROJ-123"
    - component: "Testing"
      effort: 40
      technique: "percentage"
      percentage: 0.25

  risks:
    - risk: "API complexity higher than expected"
      probability: 0.3
      impact_hours: 30
    - risk: "Design changes during development"
      probability: 0.2
      impact_hours: 20

  historical_comparison:
    similar_item: "Admin Dashboard"
    actual_effort: 180
    adjustment_factor: 0.9

yaml

estimation:
  version: "1.0"
  date: "2025-01-15"
  item: "User Dashboard Feature"
  estimator: "estimation-analyst"
  technique: "three_point"

  scope:
    description: "Interactive user dashboard with analytics"
    in_scope:
      - "Dashboard UI components"
      - "Data visualization"
      - "User preferences"
    out_of_scope:
      - "Backend analytics engine"
      - "Real-time updates"
    assumptions:
      - "API endpoints available"
      - "Design mockups complete"

  estimates:
    effort:
      optimistic: 120
      most_likely: 160
      pessimistic: 240
      expected: 166
      unit: "hours"
      std_deviation: 20
    duration:
      optimistic: 3
      most_likely: 4
      pessimistic: 6
      expected: 4.2
      unit: "weeks"
    cost:
      expected: 24000
      range_low: 18000
      range_high: 36000
      currency: "USD"

  confidence: 0.80
  contingency: 0.20

  breakdown:
    - component: "UI Components"
      effort: 60
      technique: "parametric"
      rate: "15h/component"
      quantity: 4
    - component: "Data Integration"
      effort: 40
      technique: "analogous"
      reference: "PROJ-123"
    - component: "Testing"
      effort: 40
      technique: "percentage"
      percentage: 0.25

  risks:
    - risk: "API complexity higher than expected"
      probability: 0.3
      impact_hours: 30
    - risk: "Design changes during development"
      probability: 0.2
      impact_hours: 20

  historical_comparison:
    similar_item: "Admin Dashboard"
    actual_effort: 180
    adjustment_factor: 0.9

Breakdown Table

拆分表格

markdown

undefined

markdown

undefined

Effort Breakdown

工作量拆分

Component	Technique	Estimate	Contingency	Total
UI Components	Parametric	60h	12h	72h
Data Integration	Analogous	40h	10h	50h
Testing	% of Dev	40h	8h	48h
Documentation	Parametric	16h	4h	20h
Subtotal		156h	34h	190h
Management Overhead	10%			19h
Total				209h

undefined

组件	估算技术	估算值	应急储备	总计
UI组件	参数法	60h	12h	72h
数据集成	类比法	40h	10h	50h
测试	开发工作量占比	40h	8h	48h
文档	参数法	16h	4h	20h
小计		156h	34h	190h
管理 overhead	10%			19h
总计				209h

undefined

Estimation Accuracy

估算准确率

Cone of Uncertainty

不确定性锥

Estimate accuracy improves as project progresses:

Phase	Accuracy Range
Initial concept	0.25x - 4x
Approved project	0.5x - 2x
Requirements complete	0.67x - 1.5x
Detailed design	0.8x - 1.25x
Code complete	0.9x - 1.1x

Implication: Early estimates need wider ranges; refine as knowledge grows.

估算准确率会随着项目推进逐步提升：

阶段	准确率范围
初始概念阶段	0.25x - 4x
项目获批阶段	0.5x - 2x
需求完成阶段	0.67x - 1.5x
详细设计阶段	0.8x - 1.25x
编码完成阶段	0.9x - 1.1x

含义： 早期估算的区间需要更宽，随着信息完善逐步细化估算。

Estimation Levels

估算等级

Level	Accuracy	When Used
ROM (Rough Order of Magnitude)	-25% to +75%	Initial budgeting
Budget	-10% to +25%	Project approval
Definitive	-5% to +10%	Execution baseline

等级	准确率	使用场景
ROM（粗略量级估算）	-25% 到 +75%	初步预算
预算级	-10% 到 +25%	项目审批
确定性级	-5% 到 +10%	执行基线

Common Pitfalls

常见误区

Pitfall	Prevention
Single-point estimates	Always provide ranges
Optimism bias	Use historical data, add contingency
Anchoring	Estimate before seeing others' estimates
Planning fallacy	Include realistic overhead and risks
Scope creep	Document assumptions, re-estimate on changes
Precision theater	Match precision to actual knowledge

误区	预防方法
单点估算	始终输出区间值
乐观偏差	使用历史数据，添加应急储备
锚定效应	在看到他人估算结果前先给出自己的判断
规划谬误	计入合理的overhead和风险
范围蔓延	记录假设条件，变更时重新估算
过度精确	精度和当前掌握的信息匹配即可