I am working on a dashboard using Shiny for Python and Plotly Express. I am trying to create a Gantt chart (using px.timeline) to visualize the operating periods of different boilers (ON/OFF states).

The Goal: I want to display a horizontal bar chart where the X-axis represents the timeline (Date and Time) and the bars represent the intervals where a boiler was active ("ON").

The Problem: The X-axis on the chart is not rendering the dates and times correctly. Even though I am passing datetime objects and setting the tickformat, the axis labels seem incorrect (or are not respecting the format).

My Data Transformation: I created a helper function create_gantt_data to transform my time-series data (minute-by-minute snapshots) into a DataFrame of intervals with fecha_inicio (start) and fecha_fin (end). This logic seems to work fine and returns the correct intervals:

def prepare_dataframe(df):
    # Map original CSV column names to internal aliases for easier access
    df.rename(columns={
        'Bomba Calor - Temperatura de Aire (°C)': 'temp_aire',
        'Bomba Calor - Temperatura Entrada (°C)': 'temp_entrada',
        'Bomba Calor - Temperatura Salida (°C)' : 'temp_salida',
        'Bomba Calor - Estado Caldera 2 (estado)':'estado_caldera2',
        'Bomba Calor - Estado Caldera 1 (estado)': 'estado_caldera1',
        'Bomba Calor - Estado Bomba de Calor (estado)': 'estado_bomba_calor'
    }, inplace=True, errors='raise') # Using 'raise' to enforce strict checking (errors if columns are missing)

    if ('timestamp' in df.columns):
        df['timestamp'] = pd.to_datetime(df['timestamp'], format="%d-%m-%y %H:%M").astype("datetime64[s]")
        # Set 'timestamp' as the index to enable efficient time-based filtering
        df.set_index('timestamp', inplace=True)
        df.sort_index(inplace=True)   # Sort the index chronologically (required for .loc slicing)
        
    elif ('Fecha' in df.columns and 'Hora' in df.columns):
        # If Date and Time come in separate columns, combine them
        df['timestamp'] = pd.to_datetime(df['Fecha'] + ' ' + df['Hora'], format="%d-%m-%y %H:%M").astype("datetime64[s]")
        # Set 'timestamp' as the index to enable efficient time-based filtering
        df.set_index('timestamp', inplace=True)
        df.sort_index(inplace=True)  # Sort the index chronologically (required for .loc slicing)

    final_columns = [
        'temp_aire', 'temp_entrada', 'temp_salida', 
        'estado_caldera2', 'estado_caldera1', 'estado_bomba_calor'
    ]
    
    existing_columns = []
    for col in final_columns:
        if col in df.columns:
            existing_columns.append(col)
    
    return df[existing_columns]

def create_gantt_data(df_recorte):
    gantt_rows = []
    
    for col in ['estado_caldera1', 'estado_caldera2', 'estado_bomba_calor']:

        serie = df_recorte[col]
        
        # 1. Detect state changes
        cambios = serie.diff()
        
        # 2. Identify turn-on and turn-off timestamps based on diff
        turn_on = serie[cambios == 1.0].index
        turn_off = serie[cambios == -1.0].index
        
        # 3. Handle Edge Cases
        # If it starts ON (diff misses this because there is no previous value)
        if serie.iloc[0] == 1.0:
            # Prepend the first timestamp to turn_on
            turn_on = serie.index[:1].union(turn_on)
            
        # If it ends ON (it never turned off within the selected range)
        if serie.iloc[-1] == 1.0:
            # Append the last timestamp to turn_off
            turn_off = turn_off.union(serie.index[-1:])
        
        # 4. Create the intervals
        # zip() pairs the first ON with the corresponding first OFF, etc.
        for start, end in zip(turn_on, turn_off):
            gantt_rows.append({
                'caldera': col, 
                'fecha_inicio': start,
                'fecha_fin': end
            })
    
    return pd.DataFrame(gantt_rows)

Renders a timeline chart (Gantt style) to visualize the ON intervals of each boiler.

with ui.card(style="margin-top: 20px;"):
    ui.tags.h4("Gráfico de Encendido de Calderas")
    @render_plotly
    def show_calderas():
        data = data_for_timeline()
        fig = px.timeline(
            data_frame=data,
            x_start='fecha_inicio',
            height=500,
            x_end='fecha_fin',
            template="simple_white",
            y="caldera",
            color="caldera",
            title="Tiempos de Encendido por Caldera",
            labels={
                "fecha_inicio": "Inicio de Operación",
                "fecha_fin": "Fin de Operación",
                "caldera": "Calderas"
            }
        )

        fig.update_layout(
        xaxis_title="Fecha y Hora", 
        yaxis_title="Calderas", 
        xaxis=dict(tickformat="%d-%m-%Y %H:%M")
    )
        
        fig.update_xaxes(
            tickangle=60,
            nticks=20,
        )
        return fig

example of csv please note that I am dynamically converting to timestamps:

Fecha,Hora,Bomba Calor - Temperatura de Aire (°C),Bomba Calor - Temperatura Entrada (°C),Bomba Calor - Temperatura Salida (°C),Bomba Calor - Estado Caldera 2 (estado),Bomba Calor - Estado Caldera 1 (estado),Bomba Calor - Estado Bomba de Calor (estado)
04-10-25,00:01,22.2,63.4,63.4,0.0,1.0,0.0
04-10-25,00:11,21.9,61.8,61.7,0.0,1.0,1.0
04-10-25,00:21,21.7,60.3,60.3,1.0,1.0,0.0
04-10-25,00:30,22.2,63.4,63.4,1.0,0.0,0.0
04-10-25,00:41,21.9,61.8,61.7,1.0,0.0,1.0
04-10-25,00:51,21.7,60.3,60.3,0.0,1.0,0.0
04-10-25,00:55,22.2,63.4,63.4,0.0,1.0,1.0
04-10-25,01:03,21.9,61.8,61.7,0.0,0.0,1.0
04-10-25,01:10,21.7,60.3,60.3,0.0,0.0,1.0