Related Articles

Related Articles

Data Science Apps Using Streamlit
  • Last Updated : 05 Sep, 2020

Data visualization is one of the most important steps of data analysis. It is the way to convey your research and findings of data (set) through interactive plots and charts. There are many libraries that are available for data visualization like matplotlib, seaborn, etc. which allows us to visualize a large variety of charts and plots but these libraries do not offer any functionalities to deploy them in the form of a web page or web app.

Streamlit is an open-source Python library that makes it easy to build beautiful custom web-apps for machine learning and data science. In this post we will build a small demo application in streamlit but first, we need to get an idea about some important function that we are going to use

Important functions:

  • Streamlit.title ():  This function allows you to add the title of the app.
  • Streamlit.header()/ Streamlit.subheader(): These functions are used to set header/sub-header of a section. Markdown is also supported in these function.
  • Streamlit.write(): This function is used to add anything to a web app from formatted string to charts in matplotlib figure, Altair charts, plotly figure, data frame, Keras model, etc.
  • This function is used to display maps in the web app. However, it requires the values of latitude and longitude adn these values should not be null/NA.
  • There are also common UI widgets available in streamlit library such as streamlit.button(), streamlit.checkbox(),, etc.

Caching and Performance: 

When you mark a function with streamlit.cache() annotations, it tells Streamlit that whenever the function is called it should check three things:

  • The name of the function
  • The actual code that makes up the body of the function.
  • The input parameters that you called the function with.

streamlit.cache() has following important arguments:

  • func: The function that we are going to cache. Streamlit hashes the function and its dependent code.
  • persist: This argument is used to persist the cache in the browser


  • First, we need to install streamlit into our environment, we can do it by using pip install. For this post, we will be using UK accident data, which can be downloaded from here
pip install streamlit
Building wheels for collected packages: blinker, validators, watchdog, pathtools, toolz, pyrsistent, pandocfilters
  Building wheel for blinker ( ... done
  Created wheel for blinker: filename=blinker-1.4-py3-none-any.whl size=13455 sha256=c16fdce1fa132ddc8aa6bb1551c299f64b0be90e1535383a52e3b3239e77d69c
  Stored in directory: c:\users\prashant\appdata\local\pip\cache\wheels\22\f5\18\df711b66eb25b21325c132757d4314db9ac5e8dabeaf196eab
  Building wheel for validators ( ... done
  Created wheel for validators: filename=validators-0.16.0-py3-none-any.whl size=18413 sha256=89b31cd7457f24401a146d26a8a2232d5a3e8a8ae7f1a0991dc67edbfe068a24
  Stored in directory: c:\users\prashant\appdata\local\pip\cache\wheels\38\f6\3c\5b82677e8a1b2ca411c2c10ac379dd894207329736f1ec9047
  Building wheel for watchdog ( ... done
  Created wheel for watchdog: filename=watchdog-0.10.3-py3-none-any.whl size=73875 sha256=51878f9c2c1adbe7f6d8ab6b9239c28b34990d79ee930ed1741cfa7dee0d25a6
  Stored in directory: c:\users\prashant\appdata\local\pip\cache\wheels\27\21\35\9d1e531f9de5335147dbef07e9cc99d312525ba128a93d1225
  Building wheel for pathtools ( ... done
  Created wheel for pathtools: filename=pathtools-0.1.2-py3-none-any.whl size=8790 sha256=224f8c78887059b6486a0466d5bbdd8e3fe720ebd02a29df9c6bff90445d4f79
  Stored in directory: c:\users\prashant\appdata\local\pip\cache\wheels\3e\31\09\fa59cef12cdcfecc627b3d24273699f390e71828921b2cbba2
  Building wheel for toolz ( ... done
  Created wheel for toolz: filename=toolz-0.10.0-py3-none-any.whl size=55579 sha256=815091506aeb386e2900d604ffa9803a3b879c90d8be4a5e9f8d38504f91718c
  Stored in directory: c:\users\prashant\appdata\local\pip\cache\wheels\e2\83\7c\248063997a4f9ff6bf145822e620e8c37117a6b4c765584077
  Building wheel for pyrsistent ( ... done
  Created wheel for pyrsistent: filename=pyrsistent-0.16.0-cp37-cp37m-win_amd64.whl size=70547 sha256=25abb43e1764a4266afe8485ba9476e4dc560069a06fc13cee042de747b2e750
  Stored in directory: c:\users\prashant\appdata\local\pip\cache\wheels\22\52\11\f0920f95c23ed7d2d0b05f2b7b2f4509e87a20cfe8ea43d987
  Building wheel for pandocfilters ( ... done
  Created wheel for pandocfilters: filename=pandocfilters-1.4.2-py3-none-any.whl size=7860 sha256=138f70f538c3b075a034bc0f1feee6c2868d8b16216d4ecf74987a9c71561f2e
  Stored in directory: c:\users\prashant\appdata\local\pip\cache\wheels\63\99\01\9fe785b86d1e091a6b2a61e06ddb3d8eb1bc9acae5933d4740
Successfully built blinker validators watchdog pathtools toolz pyrsistent pandocfilters
Installing collected packages: toml, urllib3, docutils, jmespath, six, python-dateutil, botocore, blinker, decorator, validators, enum-compat, s3transfer, boto3, pathtools, watchdog, numpy, pyarrow, click, MarkupSafe, jinja2, ipython-genutils, traitlets, tornado, pywin32, jupyter-core, pyzmq, jupyter-client, wcwidth, prompt-toolkit, backcall, pickleshare, pygments, parso, jedi, colorama, ipython, ipykernel, pyrsistent, zipp, importlib-metadata, attrs, jsonschema, nbformat, pywinpty, terminado, testpath, pandocfilters, entrypoints, webencodings, pyparsing, packaging, bleach, mistune, defusedxml, nbconvert, Send2Trash, prometheus-client, notebook, widgetsnbextension, ipywidgets, pydeck, pytz, pandas, toolz, altair, pillow, tzlocal, protobuf, chardet, idna, requests, base58, cachetools, astor, streamlit
Successfully installed MarkupSafe-1.1.1 Send2Trash-1.5.0 altair-4.1.0 astor-0.8.1 attrs-19.3.0 backcall-0.2.0 base58-2.0.1 bleach-3.1.5 blinker-1.4 boto3-1.14.28 botocore-1.17.28 cachetools-4.1.1 chardet-3.0.4 click-7.1.2 colorama-0.4.3 decorator-4.4.2 defusedxml-0.6.0 docutils-0.15.2 entrypoints-0.3 enum-compat-0.0.3 idna-2.10 importlib-metadata-1.7.0 ipykernel-5.3.4 ipython-7.16.1 ipython-genutils-0.2.0 ipywidgets-7.5.1 jedi-0.17.2 jinja2-2.11.2 jmespath-0.10.0 jsonschema-3.2.0 jupyter-client-6.1.6 jupyter-core-4.6.3 mistune-0.8.4 nbconvert-5.6.1 nbformat-5.0.7 notebook-6.0.3 numpy-1.19.1 packaging-20.4 pandas-1.0.5 pandocfilters-1.4.2 parso-0.7.1 pathtools-0.1.2 pickleshare-0.7.5 pillow-7.2.0 prometheus-client-0.8.0 prompt-toolkit-3.0.5 protobuf-3.12.2 pyarrow-1.0.0 
pydeck-0.4.0 pygments-2.6.1 pyparsing-2.4.7 pyrsistent-0.16.0 python-dateutil-2.8.1
 pytz-2020.1 pywin32-228 pywinpty-0.5.7 pyzmq-19.0.1 requests-2.24.0 s3transfer-0.3.3 six-1.15.0 streamlit-0.64.0 
terminado-0.8.3 testpath-0.4.4 toml-0.10.1 toolz-0.10.0 tornado-6.0.4 traitlets-4.3.3 tzlocal-2.1 urllib3-1.25.10 
validators-0.16.0 watchdog-0.10.3 wcwidth-0.2.5 webencodings-0.5.1 widgetsnbextension-3.5.1 zipp-3.1.0
  • First, we need to import the modules and load the data, we will be using to streamlit cache function to cache our data, so we don’t need to load the data again when we rerun again. After that, we use that data to plot different plot and maps. Below is the full code for the file.






# import the required modules
import streamlit as st
import pandas as pd
import numpy as np
import pydeck as pdk
import as px
# Dataset we need to import
# Add title and subtitle of the map.
st.title("Accidents in United Kingdom")
st.markdown("This app analyzes accident data in United Kingdom from 2012-2014")
Here, we define load_data function,
to prevent loading the data everytime we made some changes in the dataset.
We use streamlit's cache notation.
@st.cache(persist = True)
def load_data(nrows):
      # parse date and time columns as date and time
    data = pd.read_csv(DATA_URL, nrows = nrows, parse_dates =[['Date', 'Time']])
    # Drop N / A values in latitude and longitude. soit does not face problem when we use maps
    data.dropna(subset =['Latitude', 'Longitude'], inplace = True)
    lowercase = lambda x: str(x).lower()
    data.rename(lowercase, axis ="columns", inplace = True)
    data.rename(columns ={"date_time": "date / time"}, inplace = True)
    return data
# load first 10000 rows
data = load_data(10000)
# Plot : 1
# plot a streamlit map for accident locations.
st.header("Where are the most people casualties in accidents in UK?")
# plot the slider that selects number of person died
casualties = st.slider("Number of persons died", 1, int(data["number_of_casualties"].max()))"number_of_casualties >= @casualties")[["latitude", "longitude"]].dropna(how ="any"))
# Plot : 2
# plot a pydeck 3D map for the number of accident's happen between an hour interval
st.header("How many accidents occur during a given time of day?")
hour = st.slider("Hour to look at", 0, 23)
original_data = data
data = data[data['date / time'].dt.hour == hour]
st.markdown("Vehicle collisions between % i:00 and % i:00" % (hour, (hour + 1) % 24))
midpoint = (np.average(data["latitude"]), np.average(data["longitude"]))
    map_style ="mapbox://styles / mapbox / light-v9",
    initial_view_state ={
        "latitude": midpoint[0],
        "longitude": midpoint[1],
        "zoom": 11,
        "pitch": 50,
    layers =[
        data = data[['date / time', 'latitude', 'longitude']],
        get_position =["longitude", "latitude"],
        auto_highlight = True,
        radius = 100,
        extruded = True,
        pickable = True,
        elevation_scale = 4,
        elevation_range =[0, 1000],
# Plot : 3
# plot a histogram for minute of the hour atwhich accident happen
st.subheader("Breakdown by minute between % i:00 and % i:00" % (hour, (hour + 1) % 24))
filtered = data[
    (data['date / time'].dt.hour >= hour) & (data['date / time'].dt.hour < (hour + 1))
hist = np.histogram(filtered['date / time'].dt.minute, bins = 60, range =(0, 60))[0]
chart_data = pd.DataFrame({"minute": range(60), "Accidents": hist})
fig =, x ='minute', y ='Accidents', hover_data =['minute', 'Accidents'], height = 400)
# The code below uses checkbox to show raw data
st.header("Condition of Road at the time of Accidents")
select = st.selectbox('Weather ', ['Dry', 'Wet / Damp', 'Frost / ice', 'Snow', 'Flood (Over 3cm of water)'])
if select == 'Dry':
    st.write(original_data[original_data['road_surface_conditions']=="Dry"][["weather_conditions", "light_conditions", "speed_limit", "number_of_casualties"]].sort_values(by =['number_of_casualties'], ascending = False).dropna(how ="any"))
elif select == 'Wet / Damp':
    st.write(original_data[original_data['road_surface_conditions']=="Wet / Damp"][["weather_conditions", "light_conditions", "speed_limit", "number_of_casualties"]].sort_values(by =['number_of_casualties'], ascending = False).dropna(how ="any"))
elif select == 'Frost / ice':
    st.write(original_data[original_data['road_surface_conditions']=="Frost / ice"][["weather_conditions", "light_conditions", "speed_limit", "number_of_casualties"]].sort_values(by =['number_of_casualties'], ascending = False).dropna(how ="any"))
elif select == 'Snow':
    st.write(original_data[original_data['road_surface_conditions']=="Snow"][["weather_conditions", "light_conditions", "speed_limit", "number_of_casualties"]].sort_values(by =['number_of_casualties'], ascending = False).dropna(how ="any"))
    st.write(original_data[original_data['road_surface_conditions']=="Flood (Over 3cm of water)"][["weather_conditions", "light_conditions", "speed_limit", "number_of_casualties"]].sort_values(by =['number_of_casualties'], ascending = False).dropna(how ="any"))
if st.checkbox("Show Raw Data", False):
    st.subheader('Raw Data')


  • Now, save this code as “” and run this command as: streamlit run This will open a browser window and load the app. Let’s look at some screenshot from the app.



My Personal Notes arrow_drop_up
Recommended Articles
Page :