代写MUSA 650 Final Assignment: Remote Sensing Solution for Urban Planning代写Python语言

2025-04-01 代写MUSA 650 Final Assignment: Remote Sensing Solution for Urban Planning代写Python语言

MUSA 650 Final Assignment: Remote Sensing Solution for Urban Planning

Assignment Overview

For this final assignment in MUSA 650, you will design and implement a machine learning-based remote sensing solution to address a specific problem in urban planning.  Your goal is to clearly identify the problem, define your target users, and develop a solution that uses Python-based remote sensing and machine learning techniques.  This project will evaluate both your technical approach (50%) and a polished, five-minute lightning talk (50%) geared toward professional audiences.

Objectives

   ● Identify an urban planning problem and define your target user.

   ● Develop a solution using remote sensing data and machine learning techniques in Python.

   ● Provide a comprehensive analysis of the solution, including data sources, model architecture, preprocessing steps, feature engineering, and validation.

   ● Evaluate the solution's quantitative performance and discuss qualitative aspects like constraints, limitations, and improvements.

   ● Present a professional-grade, portfolio-ready summary of your work.

Problem Scope

1.  Problem Identification

   ● Identify a specific urban planning problem to address.  Possible topics include:

      ○ Urban sprawl analysis

      ○ Flood identification

      ○ Heat risk mapping

      ○ Green corridor identification

   ● Explain why this is a problem, its impacts, and its relevance to urban planning.

2.  Target User(s)

   ● Define a hypothetical user such as a national think tank, a city or regional government, or a research institution (e.g., USGS).

   ● Describe why this user would value your solution and the insights they hope to gain.

Solution Development

3.  Data and Model

   ● Data Source: Specify the data source(s) you will use, for example:

      ○ NLCD (National Land Cover Database) for land cover classification

      ○ Sentinel-1 or Sentinel-2 for radar and optical data

      ○ Landsat for multi-decadal change detection

      ○ Proprietary datasets that are accessible through open access licenses

   ● Model Architecture: Describe your model architecture, including:

      ○ Algorithms and libraries you plan to use (e.g., Google Earth Engine, TorchGeo, Keras).

      ○ Preprocessing steps and feature engineering.

      ○ Alternative approaches you considered and your rationale for the final model selection.

   ● Scaling Considerations: Discuss hardware and software considerations, such as:

      ○ Data Augmentation: When and why data augmentation may be necessary.

      ○ Cloud Computing and GPUs: Whether cloud computing (e.g., Google Cloud Platform, AWS) or GPUs are required to handle large datasets or speed up training.

      ○ Training Constraints: Describe any constraints on model training (e.g., speed limitations with local machines versus cloud environments) and how they influenced your model selection.

4.  Relevant Research

   ● Summarize relevant research papers that informed your approach in brief paragraphs.

   ● Highlight key insights and how they influenced your design.

Evaluation

5.  Quantitative and Qualitative Analysis

   ● Evaluation Metrics: Describe and justify the evaluation metrics you used (e.g., accuracy, precision, recall, F1-score, Intersection over Union).

   ● Quantitative Evaluation: Report your model’s performance based on these metrics.

   ● Qualitative Analysis: Describe strengths and limitations of your solution.  Consider:

      ○ Labeled Data Availability: Was there enough labeled data, and did this affect the model’s accuracy?

      ○ Data Augmentation: Discuss if data augmentation was used to improve performance.

      ○ Computation Requirements: Describe how computation requirements (e.g., using a local machine vs. cloud computing) impacted your solution.

      ○ Scalability: Address the feasibility of scaling this solution and the resources needed.

6.  Future Steps

   ● Identify steps for improvement or expansion, such as enhancing model architecture, integrating additional data sources, or adjusting parameters.

Lightning Talk Presentation

Requirements

   ● Prepare a five-minute lightning talk that professionally summarizes your problem, solution, and key findings.

   ● Your presentation should be polished and concise, suitable for inclusion in a professional portfolio.

   ● Suggested formats include:

      ○ A video demonstration

      ○ A blog post

      ○ A LinkedIn article or post

      ○ A slide deck with speaker notes

   ● Your talk should clearly define the problem, outline your approach, present evaluation results, and discuss key insights or takeaways.

Submission and Grading

   ● Code and Analysis (50%): Evaluation based on the technical depth, workflow, and rigor of your code, analysis, and evaluation.

   ● Lightning Talk (50%): Grading will focus on clarity, professionalism, presentation quality, and alignment with a professional portfolio.

Make sure your submission is well-documented and easy to follow for any potential viewers or evaluators.  Best of luck, and we look forward to seeing your innovative solutions!