Organic Slider Method?

Using Streamlit
2

I solved it with ‘st.experimental_rerun()’.

To prevent the slider from being created twice, I entered the session value in the slider’s value, put the calculated value in the session, and ran ‘st.experimental_rerun()’.

However, it adds the calculation once when reloading the page, so I added another session that doesn’t recalculate if the settings haven’t changed.

Here’s the code for the slider part


    if len(recommend_stocks) > 0:
        optimal_weights = rp.get_portfolio_optimal(recommend_stocks)
        column_count = 5
        # print(f"-------------------")
        # print(f"optimal_weights: {optimal_weights}")

        if 'optimal_weights' not in st.session_state:
            st.session_state['optimal_weights'] = optimal_weights

        if optimal_weights != st.session_state['optimal_weights']:
            st.session_state['optimal_weights'] = optimal_weights
            st.session_state['guide_weight'] = optimal_weights

        #.
        if 'guide_weight' not in st.session_state:
            st.session_state['guide_weight'] = optimal_weights

        guide_weight = st.session_state['guide_weight']

        # Create a reweighting slider in the UI logic.
        col_weight_reco = st.columns(column_count)
        for idx, stock in enumerate(recommend_stocks):
            stock_name = option_info['stock_match'][stock]
            recommend_stock_names.append(stock_name)
            stock_link = f "[{stock_name}](https://rank.newsystock.com/NewsyRank/StockAnalysis/SA_Diagnosis.aspx?StockCode={stock}&Univ=KF-ALL)"

            weight_value = st.session_state['guide_weight'][idx] if idx < len(st.session_state['guide_weight']) else math.floor(
                1 / len(recommend_stocks) * 10000) / 100

            recommend_weight = col_weight_reco[idx % column_count].slider(stock_link, min_value=0.00, max_value=100.00, value=weight_value, step=1.00)
            recommend_weights.append(recommend_weight)

            if (idx + 1) % column_count == 0:
                # if there are 5 columns, create a new column.
                col_weight_reco = st.columns(column_count)

        # st.session_state['recommend_weight'] = recommend_weights
        if recommend_weights != st.session_state['guide_weight']:
            # print(f"guide : {st.session_state['guide_weight']}")
            # print(f"recom : {recommend_weights}")

            # guide and recom different indexes.(
            diff_idx = [idx for idx, (g, r) in enumerate(zip(guide_weight, recommend_weights)) if g != r]
            # print(diff_idx)

            # Calculate remaining weight, excluding that index.
            remain_weight = 100 - sum([recommend_weights[idx] for idx in diff_idx])
            origin_sum = sum([optimal_weights[idx] for idx in range(len(recommend_weights)) if idx not in diff_idx])

            # apply that index with the changed weight, and change the weights of other indexes to match the optimal_weights weight.
            for idx in range(len(recommend_weights)):
                if idx in diff_idx:
                    pass
                else:
                    recommend_weights[idx] = round(optimal_weights[idx] / origin_sum * remain_weight, 2)
            st.session_state['guide_weight'] = recommend_weights
            # print(f"change : {recommend_weights}, sum : {sum(recommend_weights)}")

            st.experimental_rerun()

The results of the behavior can be seen on the next page.

https://genportview.com/portfolio-manager

Behavior screen

streamlit-🎨_젠포트뷰-2023-05-25-14-05-14