You are conducting a complex web task that requires information from the web to answer correctly. Directly navigating the web environment to provide a final answer cannot always yield the correct result. Therefore, you need to decompose the task into two decoupled parts to complete it successfully.
The two parts are the navigation part and the analysis part.
The navigation part involves visiting all pages that contain the data needed to solve the task. The observation, the accessibility tree of full web page, at each step will be recorded during navigation.
The analysis part involves extracting information from the observations and writing code to provide the final answer. Note that the extracted information processed during analysis part may be imperfect, which means they may include unnecessary data or not in correct format, you need to make sure the analysis code can be robust to handle such cases.
Another important consideration is to simplify the navigation, as it is a more challenging task. Ignore constraints such as ranges or filters in the navigation objective. Instead, include such constraints in the analysis part to be handled later.
Given the original complex user task and some tips for using the target website, decompose it into these two parts following this approach. Your output must follow this format with exact the same headers:

### Part 1 – Navigation


### Part 2 – Analysis



In addition, below are some decomposition examples for your reference:

Example 1:

User task  
“List the average rating for every movie genre, using only titles released between 2015 and 2024. Output: ‘Drama : 8.1, Comedy : 7.4, …’”

### Part 1 – Navigation  
Go to the pages which include each film’s genre, release year, and numeric user rating. Do not go to each film detail page if all the information is available in film listing page.

### Part 2 – Analysis  
Filter and only keep only films released 2015-2024. Compute the average rating per genre and show them as ‘Drama : X.X, Comedy : Y.Y, …’.

Example 2:

User task  
“Among products tagged ‘wireless earbuds’, count how many cost below $50, $50-$99, and $100+. Return: ‘<50 : __, 50-99 : __, 100+ : __’.”

### Part 1 – Navigation  
Visit the pages containing product title and price information for “wireless earbuds” products. Do not go to each product detail page if all the information is available in product listing page.

### Part 2 – Analysis  
Group the collected items by price brackets  < $50, $50-$99, $100+.  Count how many fall into each bracket and output the counts in the following format: ‘<50 : __, 50-99 : __, 100+ : __’

Example 3:

User task  
“In the travel forum, among the 200 latest hotel reviews, how many mention ‘noise’ or ‘quiet’ in the text? Give two numbers: noisy_count, quiet_count.”

### Part 1 – Navigation  
Navigate to the pages including the text body of the 200 most recent hotel reviews in the travel forum. Do not go to each review detail page if all the information is available in review listing page.

### Part 2 – Analysis  
Only keep the 200 most recent reviews. Search each saved review for the words “noise”, “noisy” (noisy_count) and “quiet”. Return two integers: noisy_count and quiet_count.
