SectorScan Part 2: Building a Financial Intelligence App w/ Streamlit

Hey there, welcome to the second part of SectorScan development recipe! If you haven’t checked out the first part of this recipe, please check it here. As I mentioned before, in this part, we’ll elevate everything we’ve accomplished by seamlessly integrating it into Streamlit. This will transform our work into a fully functional, interactive app. Get ready to see SectorScan come to life as a powerful tool for financial analysis and visualization! To give you an overview of what SectorScan will look like, you can visit this link.

Note: to follow along this recipe, you can’t use Google Colab anymore. Please use Visual Studio Code or any other local IDE of your choice.

Getting familiar with Streamlit

Why do we choose Streamlit to build SectorScan? Because it has these key features:

Simple: Streamlit’s intuitive API allows you to create powerful web apps with minimal effort.
Fast: Quickly iterate on your app and see changes in real-time with Streamlit’s automatic hot reloading.
Flexible: Streamlit supports a wide range of data visualization libraries, making it easy to incorporate charts, tables, and other visualizations into your app.
Sharing: Share your apps with others by deploying them to Streamlit’s cloud platform or hosting them on your own server.

The first thing we wanna do here is to get familiar with how to create your first Streamlit app:

Let’s create a folder called sectorscan in your chosen working directory and create a file called sectorscan.py inside it.

Please make sure you’ve already installed streamlit in your local. If you’ve done that, now import it to your Python script (sectorscan.py) using this code:

import streamlit as st

# Import some other libraries that we're going to need in our code later
import pandas as pd
import requests
import altair as alt

Using this documentation as your reference, write your app using Streamlit’s simple API:
```
# Title
st.title("SectorScan")

# Content
st.write("Welcome to SectorScan!")
```
Run your app from the command line:
```
streamlit run sectorscan.py
```
That’s it! You’ve created your first Streamlit app. The previous command line you run should redirect you to the browser showing your app as follows:

If you want to stop the app, just hit CTRL + C in your command line.

In the next section, we’ll start to move everything we’ve created in the previous part to this app.

Build the SectorScan

To make your life easier, keep this documentation open while you follow along with this tutorial. Feel free to also modify my code if you find a better way to display the app’s contents!

Data retrieval function

First thing that we want to move to our app is the function to retrieve Sectors data. But certainly, there are some modifications we need to do. One thing we need to modify is the way we call the API key. There’s no way we’re going to display the API key right away in our code for our safety. Thus, we’re going to utilize streamlit secrets in our function:

Create a folder called .streamlit in your sectorscan folder, and fill it with a file named secrets.toml.
In the TOML file, create a new variable called SECTORS_KEY and paste your API key there:
```
SECTORS_KEY="Your API Key"
```
Access the secrets from your Python script using st.secrets:
```
api_key = st.secrets["SECTORS_KEY"]
```

Next, to handle error from API request, previously we use raise Exception. Since we’re creating an app, we need to let the user knows if something went wrong, but in the same time, not being too specific. Thus, we’ll use st.error() to notice the user about the error:

st.error("Error: Something went wrong. Please reload the app.")

We’ll also add a st.cache_data decorator to our function to make use of cache on our function calls. Cache is a mechanism for temporarily storing frequently accessed data to improve performance and reduce redundant computations.

So the final function looks like this:

@st.cache_data
def fetch_data(url):
  api_key = st.secrets["SECTORS_KEY"]

  headers = {
      "Authorization": api_key
  }

  response = requests.get(url, headers = headers)

  if response.status_code == 200:
      return response.json()
  else:
      # Handle error
      st.error("Error: Something went wrong. Please reload the app.")

You can test the function by retrieving the sectors data and display it as a text first using st.write():

url = "https://api.sectors.app/v1/subsectors/"
data = fetch_data(url)
sectors = [item['subsector'] for item in data]
sectors.sort() # sort the list based on the sector's name to make it tidier
st.write(sectors)

Notice that your app won’t be changed when you modify your code:

To automatically reload your app whenever the source code change, click the Always rerun button on the top right of the page. But if you only want to reload the app when you ask the app to do so, click the Rerun button instead. If the function works, your app should look like this after you click the button (I’ve removed the welcome text to make the app tidier):

Ok since the function works already, remove the st.write() and let’s continue to bring more content to our app!

Sectors filter component

Since we want to compare data from different sectors in this app, we want to make sure that we provide a component for the user to choose which sectors they want to compare. Thus, we’re going to provide a multiselect filter that displays all the available sectors. We’ll use st.multiselect() for this:

options = st.multiselect(
    label="Filter Sectors",
    options=sectors,
    default=sectors[0:3],
    max_selections=5
)

We use options variable to contain the selected sectors to be used across our code later. And to ensure the visualizations remain clear and easy to read, we limit the maximum selections to 5 sectors. When your app reloads, it should show the multiselect filter as follows:

Notice that the options are in ..-.. format, so we’ll utilize format_func parameter of the st.multiselect() to display the options in a prettier way:

def format_option(option):
    return option.replace("-", " ").title() # change - to space and use title case

options = st.multiselect(
    label="Filter Sectors",
    options=sectors,
    default=sectors[0:3],
    format_func=format_option,
    max_selections=5
)

Ok, now we have a better filter to be displayed:

If you read the documentation of st.multiselect(), there’s no option to disable the clear options button, thus, we have to make sure that we handle the case when user clears the options (no sector is selected). We’ll use st.warning() to warn the user about this:

if len(options) == 0:
    st.warning("Please select at least one sector!")
else:
    st.write("Visualization") # later on change it to the visualization codes

Now if you clear the options you’ll see this warning:

Market cap section

In the market cap section, we’re going to display two rows:

The first row will be divided into two columns to display the total market cap and historical market cap visualization
The second row will show the historical market cap change visualization.

But first, of course we need to move all the code to retrieve the market cap data to the sectorscan.py. Copy paste the last chunk of code of this section.

Since now we’re retrieving the data based on the selected sectors, we need to modify the code a bit. Instead of using range(3), we’ll change it to our options variable:

df_mc_curr = pd.DataFrame()
df_mc_hist = pd.DataFrame()
df_mc_change = pd.DataFrame()

for i in options:
    url = f"https://api.sectors.app/v1/subsector/report/{i}/?sections=market_cap"
    market_cap = fetch_data(url)
    ... # the rest of the code

First row of the market cap section

To process the first row, don’t forget to also move the following codes to sectorscan.py:

Code to produce the final data frame of df_mc_curr and the mc_curr_chart code in this section
Code to produce the final data frame of df_mc_hist and the mc_hist_chart code in this section.

To display the first row as two columns, we’ll use st.columns(). And to display the altair charts, we’ll use st.altair_chart():

# Market cap section title
st.subheader("Market Cap")

# First row visualization
col1, col2 = st.columns(
    spec=[0.6, 0.4], # relative width of each column
    gap="large" # size of gap between column
)

# First column
with col1:
    st.altair_chart(mc_curr_chart)

# Second column
with col2:
    st.altair_chart(mc_hist_chart)

The result will be as follow:

Feel free to modify the width of the columns and the charts.

Second row of the market cap section

For the second row, let’s first move the code to produce the final data frame of df_mc_change and the code of mc_change_chart from this section to sectorscan.py.

Since the second row is not divided into columns, we can directly use st.altair_chart() to display the chart:

# Second row visualization
st.altair_chart(mc_change_chart)

The result will be as follow:

Valuation section

First let’s move the code to retrieve valuation data from the last chunk of code of this section. Don’t forget to change the range(3) to options:

df_valuation = pd.DataFrame()

for i in options:
    url = f"https://api.sectors.app/v1/subsector/report/{i}/?sections=valuation"
    valuation = fetch_data(url)
    ... # the rest of the code

In this section, we’re going to only have one visualization, but we’ll let the user choose what valuation metric they want to show. We’ll use st.selectbox() to do so:

# Valuation section title
st.subheader("Valuation")

# Valuation metric filter
option = st.selectbox(
    label="Select Valuation Metric",
    options=("Price/Book Ratio", "Price/Earning Ratio", "Price/Sales Ratio", "Price/Cash Flow Ratio")
)

The option variable will contain the selected valuation metric and we’ll use it to make our valuation_chart to be dynamic. Move the valuation_chart code from this section to sectorscan.py, but change all the Price/Book Ratio to option. Then, display it using st.altair_chart():

# valuation_chart
valuation_chart = alt.Chart(df_valuation).mark_line(
    point=True # add individual data points to the line chart
).encode(
    x=alt.X("Year:N", axis=alt.Axis(labelAngle=0)), # 0 degree of x-axis label angle
    y=alt.Y(f"{option}:Q"),
    color=alt.Color(
        "Sector:N",
        scale=alt.Scale(scheme="lightgreyred"),
        sort=alt.SortField(field="Sector", order="ascending") # sort color based on the sector's name
    ),
    tooltip=[
        alt.Tooltip("Sector:N"),
        alt.Tooltip("Year:N"),
        alt.Tooltip(f"{option}:Q", format=".2f")
    ]
).properties(
    title=f"{option} Across Sectors",
    width=900
)

st.altair_chart(valuation_chart)

Your valuation section should look like this:

If you change the selection, the chart should be updated based on it.

Top companies section

In this section, we’ll use tabs to display the visualizations. There are four tabs we want to create here:

Market Cap to display Top companies based on Market Cap
Growth to display Top companies based on Revenue Growth
Profit to display Top companies based on Profit
Revenue to display Top companies based on Revenue

Let’s first move the code to retrieve top companies data from the last chunk of code in this section to sectorscan.py. Also change the range(3) to options:

df_top_mc = pd.DataFrame()
df_top_growth = pd.DataFrame()
df_top_profit = pd.DataFrame()
df_top_revenue = pd.DataFrame()

for i in options:
    url = f"https://api.sectors.app/v1/subsector/report/{i}/?sections=companies"
    company = fetch_data(url)
    ... # the rest of the code

Also move the following codes to sectorscan.py:

Code to produce the final data frame of df_top_mc and the mc_chart code in this section
Code to produce the final data frame of df_top_growth and the growth_chart code in this section
Code to produce the final data frame of df_top_profit and the profit_chart code in this section
Code to produce the final data frame of df_top_revenue and the revenue_chart code in this section.

Then, to display all the visualizations on each of its tab, we’ll use st.tabs():

# Top companies section title
st.subheader("Top Companies")

# Top companies visualization
tab1, tab2, tab3, tab4 = st.tabs(
    tabs=["Market Cap", "Growth", "Profit", "Revenue"]
)

# First tab
with tab1:
    st.altair_chart(mc_chart)

# Second tab
with tab2:
    st.altair_chart(growth_chart)

# Third tab
with tab3:
    st.altair_chart(profit_chart)

# Fourth tab
with tab4:
    st.altair_chart(revenue_chart)

Your top companies section should look like this:

When you click on another tab’s title, it should display the chart based on the title.

(Optional) Deploy the SectorScan

Actually if you only need your app locally, you can stop right here. But if you want your app to be accessible anywhere, you need to continue to this section.

Prerequisite

Some additional prerequisites are required to be able to deploy the app:

Step by step

We’re going to follow this documentation to deploy our app:

Login and go to Github, then click the New button:
Fill in the repository details, more or less as follows:
Click the Create repository button.
Now, prepare your files:
- sectorscan.py that we’ve created throughout this recipe.
- requirements.txt with the content as follows:
  streamlit==1.35.0
Upload both of the files to the repository by clicking the Add file - Upload files button on the repository page:

Then, click on the Commit changes button. Your repository is ready to be deployed! Your repository page now should look like this:
Login to Streamlit to deploy the app.
Click the Create app button on the top right:
Click on the Yup, I have an app option.
Fill in the app detail as follows:

You should choose different app URL since I’ve used sectorscan.streamlit.app.
Click the Advance settings... link below the App URL and paste the content of secrets.toml in there:

Then, click the Save button.
Last step, click the Deploy! button and congrats, your app is ready!

Go beyond

Congratulations on completing this recipe! You’ve taken the first step in creating a powerful financial analytics tool. Now, the possibilities are endless. You can always expand and improve your app by adding more visualizations and analyses. Dive deeper into the data available from the Sectors Financial API to uncover new insights and make your app even more comprehensive and insightful. Happy coding!

Note: The full version of the code can be accessed here.

Quick Start in Python

Stock Investing and Finance

Financial Analytics w/ Streamlit

Animated Plots in R

API in Spreadsheets & BI Tools

SectorScan Part 2: Building a Financial Intelligence App w/ Streamlit

Getting familiar with Streamlit

Build the SectorScan

Data retrieval function

Sectors filter component

Market cap section

First row of the market cap section

Second row of the market cap section

Valuation section

Top companies section

(Optional) Deploy the SectorScan

Prerequisite

Step by step

Go beyond

Quick Start in Python

Stock Investing and Finance

Financial Analytics w/ Streamlit

Animated Plots in R

API in Spreadsheets & BI Tools

​Getting familiar with Streamlit

​Build the SectorScan

​Data retrieval function

​Sectors filter component

​Market cap section

​First row of the market cap section

​Second row of the market cap section

​Valuation section

​Top companies section

​(Optional) Deploy the SectorScan

​Prerequisite

​Step by step

​Go beyond

Getting familiar with Streamlit

Build the SectorScan

Data retrieval function

Sectors filter component

Market cap section

First row of the market cap section

Second row of the market cap section

Valuation section

Top companies section

(Optional) Deploy the SectorScan

Prerequisite

Step by step

Go beyond